Query builder

The Database query builder is used to construct SELECT, UPDATE, and DELETE SQL queries. For inserting new rows you must use the insert query builder and use raw query builder for running raw SQL queries.

You can get an instance of the database query builder using one of the following methods.

import Database from '@ioc:Adonis/Lucid/Database'

Database.query()

// selecting table returns the query builder instance as well
Database.from('users')

Methods/properties

Following is the list of available methods/properties available on the Query builder instance.

select

The select method allows selecting columns from the database table. You can either pass an array of columns or pass them as multiple arguments.

Database
  .from('users')
  .select('id', 'username', 'email')

Column aliases

You can define aliases for the columns using the as expression or passing an object of key-value pair.

Database
  .from('users')
  .select('id', 'email as userEmail')

Database
  .from('users')
  .select({
    id: 'id',

    // Key is alias name
    userEmail: 'email'
  })

Using sub queries

Also, you can make use of sub-queries and raw-queries for generating columns at runtime, for example, selecting the last login IP address for a user from the user_logins table.

Database
  .from('users')
  .select(
    Database
      .from('user_logins')
      .select('ip_address')
      .whereColumn('users.id', 'user_logins.user_id')
      .orderBy('id', 'desc')
      .limit(1)
      .as('last_login_ip') // 👈 This is important
  )

Using raw queries

Similar to a sub-query, you can pass an instance of the raw query as well.

Database
  .from('users')
  .select(
    Database
      .raw(`
        (select ip_address from user_logins where users.id = user_logins.user_id limit 1) as last_login_ip
      `)
  )

from

The from method is used to define the database table for the query.

Database.from('users')

The query builder also allows using derived tables by passing a sub-query or a closure (which acts like a sub-query).

Database.from((subquery) => {
  subquery
    .from('user_exams')
    .sum('marks as total')
    .groupBy('user_id')
    .as('total_marks')
}).avg('total_marks.total')

where

The where method is used to define the where clause in your SQL queries. The query builder accepts a wide range of arguments types to let you leverage the complete power of SQL.

The following example accepts the column name as the first argument and its value as the second argument.

Database
  .from('users')
  .where('username', 'virk')

You can also define SQL operators, as shown below.

Database
  .from('users')
  .where('created_at', '>', '2020-09-09')

// Using luxon to make the date
Database
  .from('users')
  .where('created_at', '>', DateTime.local().toSQLDate())

// Using like operator
Database
  .from('posts')
  .where('title', 'like', '%Adonis 101%')

Where groups

You can create where groups by passing a callback to the where method. For example:

Database
  .from('users')
  .where((query) => {
    query
      .where('username', 'virk')
      .whereNull('deleted_at')
  })
  .orWhere((query) => {
    query
      .where('email', 'virk@adonisjs.com')
      .whereNull('deleted_at')
  })

Generated SQL

SELECT * FROM "users"
  WHERE (
    "username" = ? AND "deleted_at" IS NULL
  )
  or (
    "email" = ? AND "deleted_at" IS NULL
  )

Using sub queries

The where method value can also be a sub-query.

Database
  .from('user_groups')
  .where(
    'user_id',
    Database
      .from('users')
      .select('user_id')
      .where('users.user_id', 1)
  )

Using raw queries

Similarly, you can also define a raw query.

Database
  .from('user_groups')
  .where(
    'user_id',
    Database
      .raw(`select "user_id" from "users" where "users"."user_id" = ?`, [1])
      .wrap('(', ')')
  )

where method variants

Following is the list of the where method variations and shares the same API.

Method	Description
`andWhere`	Alias for the `where` method
`orWhere`	Adds an or where clause
`whereNot`	Adds a where not clause
`orWhereNot`	Adds an or where not clause
`andWhereNot`	Alias for `whereNot`

whereColumn

The whereColumn method allows you to define a column as the value for the where clause. The method is usually helpful with queries and joins. For example:

Database
  .from('users')
  .whereColumn('updated_at', '>', 'created_at')

Database
  .from('users')
  .select(
    Database
      .from('user_logins')
      .select('ip_address')
      .whereColumn('users.id', 'user_logins.user_id') // 👈
      .orderBy('id', 'desc')
      .limit(1)
      .as('last_login_ip')
  )

whereColumn method variants

Following is the list of the whereColumn method variations and shares the same API.

Method	Description
`andWhereColumn`	Alias for the `whereColumn` method
`orWhereColumn`	Adds an or where clause
`whereNotColumn`	Adds a where not clause
`orWhereNotColumn`	Adds an or where not clause
`andWhereNotColumn`	Alias for `whereNotColumn`

whereLike

Adds a where clause with case-sensitive substring comparison on a given column with a given value.

Database
  .from('posts')
  .whereLike('title', '%Adonis 101%')

whereILike

Adds a where clause with case-insensitive substring comparison on a given column with a given value. The method generates a slightly different for each dialect to achieve the case insensitive comparison.

Database
  .from('posts')
  .whereILike('title', '%Adonis 101%')

whereIn

The whereIn method is used to define the wherein SQL clause. The method accepts the column name as the first argument and an array of values as the second argument.

Database
  .from('users')
  .whereIn('id', [1, 2, 3])

The values can also be defined for more than column. For example:

Database
  .from('users')
  .whereIn(['id', 'email'], [
    [1, 'virk@adonisjs.com']
  ])

// SQL: select * from "users" where ("id", "email") in ((?, ?))

Using sub queries

You can also compute the whereIn values using a subquery.

Database
  .from('users')
  .whereIn(
    'id',
    Database
      .from('user_logins')
      .select('user_id')
      .where('created_at', '<', '2020-09-09')
  )

For multiple columns

Database
  .from('users')
  .whereIn(
    ['id', 'email'],
    Database
      .from('accounts')
      .select('user_id', 'email')
  )

The whereIn method also accepts a callback as the 2nd argument. The callback receives an instance of the subquery that you can use to compute values as runtime.

Database
  .from('users')
  .whereIn(
    'id',
    (query) => query.from('user_logins').select('user_id')
  )

whereIn method variants

Following is the list of the whereIn method variations and shares the same API.

Method	Description
`andWhereIn`	Alias for the `whereIn` method
`orWhereIn`	Adds an or where in clause
`whereNotIn`	Adds a where not in clause
`orWhereNotIn`	Adds an or where not in clause
`andWhereNotIn`	Alias for `whereNotIn`

whereNull

The whereNull method adds a where null clause to the query.

Database
  .from('users')
  .whereNull('deleted_at')

whereNull method variants

Following is the list of the whereNull method variations and shares the same API.

Method	Description
`andWhereNull`	Alias for the `whereNull` method
`orWhereNull`	Adds an or where null clause
`whereNotNull`	Adds a where not null clause
`orWhereNotNull`	Adds an or where not null clause
`andWhereNotNull`	Alias for `whereNotNull`

whereExists

The whereExists method allows adding where constraints by checking for the existence of results on a subquery. For example: Select all users who have at least logged in once.

Database
  .from('users')
  .whereExists((query) => {
    query
      .from('user_logins')
      .whereColumn('users.id', 'user_logins.user_id')
      .limit(1)
  })

You can also pass in a sub-query or a raw query as the first argument.

Database
  .from('users')
  .whereExists(
    Database
      .from('user_logins')
      .whereColumn('users.id', 'user_logins.user_id')
      .limit(1)
  )

Database
  .from('users')
  .whereExists(
    Database.raw(
      'select * from user_logins where users.id = user_logins.user_id'
    )
  )

whereExists method variants

Following is the list of the whereExists method variations and shares the same API.

Method	Description
`andWhereExists`	Alias for the `whereExists` method
`orWhereExists`	Adds an or where exists clause
`whereNotExists`	Adds a where not exists clause
`orWhereNotExists`	Adds an or where not exists clause
`andWhereNotExists`	Alias for the `whereNotExists` method

whereBetween

The whereBetween method adds the where between clause. It accepts the column name as the first argument and an array of values as the second argument.

Database
  .from('users')
  .whereBetween('age', [18, 60])

Using sub queries

You can also use subqueries to derive the values from a different database table.

Database
  .from('users')
  .whereBetween('age', [
    Database.from('participation_rules').select('min_age'),
    Database.from('participation_rules').select('max_age'),
  ])

Using raw queries

You can also make use of raw queries for deriving values from another database table.

Database
  .from('users')
  .whereBetween('age', [
    Database.raw('(select min_age from participation_rules)'),
    Database.raw('(select max_age from participation_rules)'),
  ])

whereBetween method variants

Following is the list of the whereBetween method variations and shares the same API.

Method	Description
`andWhereBetween`	Alias for the `whereBetween` method
`orWhereBetween`	Adds an or where between clause
`whereNotBetween`	Adds a where not between clause
`orWhereNotBetween`	Adds an or where not between clause
`andWhereNotBetween`	Alias for the `whereNotBetween` method

whereRaw

You can use the whereRaw method to express conditions not covered by the existing query builder methods.

Always make sure to bind parameters and do not encode user input directly in the raw query.

❌ Encoding user values directly

Database
  .from('users')
  .whereRaw(`username = ${username}`)

✅ Using bind params

Database
  .from('users')
  .whereRaw('username = ?', [username])

You can also define the column names dynamically using ??.

Database
  .from('users')
  .whereRaw('?? = ?', ['users.username', username])

whereRaw method variants

Following is the list of the whereRaw method variations and shares the same API.

Method	Description
`andWhereRaw`	Alias for the `whereRaw` method
`orWhereRaw`	Adds an or where raw clause

whereJson

Add a where clause with an object to match the value of a JSON column inside the database.

Database
  .from('users')
  .whereJson('address', { city: 'XYZ', pincode: '110001' })

The column value can also be computed using a sub-query.

Database
  .from('users')
  .whereJson(
    'address',
    Database
      .select('address')
      .from('user_address')
      .where('address.user_id', 1)
  )

whereJson method variants

Following is the list of the whereJson method variations and shares the same API.

Method	Description
`orWhereJson`	Add a or where clause matching the value of a JSON column
`andWhereJson`	Alias for `whereJson`
`whereNotJson`	Add a where not clause against a JSON column
`orWhereNotJson`	Add a or where not clause against a JSON column
`andWhereNotJson`	Alias for `whereNotJson`

whereJsonSuperset

Add a clause where the value of the JSON column is the superset of the defined object. In the following example, the user address is stored as JSON and we find by the user by their pincode.

Database
  .from('users')
  .whereJsonSuperset('address', { pincode: '110001' })

whereJsonSuperset method variants

Following is the list of the whereJsonSuperset method variations and shares the same API.

Method	Description
`orWhereJsonSuperset`	Add a or where clause matching the value of a JSON column
`andWhereJsonSuperset`	Alias for `whereJsonSuperset`
`whereNotJsonSuperset`	Add a where not clause against a JSON column
`orWhereNotJsonSuperset`	Add a or where not clause against a JSON column
`andWhereNotJsonSuperset`	Alias for `whereNotJsonSuperset`

whereJsonSubset

Add a clause where the value of the JSON column is the subset of the defined object. In the following example, the user address is stored as JSON and we find by the user by their pincode or the city name.

Database
  .from('users')
  .whereJsonSubset('address', { pincode: '110001', city: 'XYZ' })

whereJsonSubset method variants

Following is the list of the whereJsonSubset method variations and shares the same API.

Method	Description
`orWhereJsonSubset`	Add a or where clause matching the value of a JSON column
`andWhereJsonSubset`	Alias for `whereJsonSubset`
`whereNotJsonSubset`	Add a where not clause against a JSON column
`orWhereNotJsonSubset`	Add a or where not clause against a JSON column
`andWhereNotJsonSubset`	Alias for `whereNotJsonSubset`

join

The join method allows specifying SQL joins between two tables. For example: Select the ip_address and the country columns by joining the user_logins table.

Database
  .from('users')
  .join('user_logins', 'users.id', '=', 'user_logins.user_id')
  .select('users.*')
  .select('user_logins.ip_address')
  .select('user_logins.country')

You can pass a callback as the 2nd argument to define more join constraints.

Database
  .from('users')
  .join('user_logins', (query) => {
    query
      .on('users.id', '=', 'user_logins.user_id')
      .andOnVal('user_logins.created_at', '>', '2020-10-09')
  })
  .select('users.*')
  .select('user_logins.ip_address')
  .select('user_logins.country')

To group join constraints, you can pass a callback to the on method.

Database
  .from('users')
  .join('user_logins', (query) => {
    query
      .on((subquery) => {
        subquery
          .on('users.id', '=', 'user_logins.user_id')
          .andOnVal('user_logins.created_at', '>', '2020-10-09')
      })
      .orOn((subquery) => {
        subquery
          .on('users.id', '=', 'user_logins.account_id')
          .andOnVal('user_logins.created_at', '>', '2020-10-09')
      })
  })
  .select('users.*')
  .select('user_logins.ip_address')
  .select('user_logins.country')

Output SQL

SELECT
  "users".*,
  "user_logins"."ip_address",
  "user_logins"."country"
FROM "users"
  INNER JOIN "user_logins" ON (
    "users"."id" = "user_logins"."user_id" AND "user_logins"."created_at" > ?
  )
  or (
    "users"."id" = "user_logins"."account_id" AND "user_logins"."created_at" > ?
  )

The join method uses the inner join by default, and you can use a different join using one of the following available methods.

leftJoin
leftOuterJoin
rightJoin
rightOuterJoin
fullOuterJoin
crossJoin

joinRaw

You can use the joinRaw method to express conditions not covered by the query builder standard API.

On methods

Following is the list of available on methods you can use with a join query.

Database
  .from('users')
  .joinRaw('natural full join user_logins')

onIn

Database
  .from('users')
  .join('user_logins', (query) => {
    query.onIn('user_logins.country', ['India', 'US', 'UK'])
  })

onNotIn

Database
  .from('users')
  .join('user_logins', (query) => {
    query.onNotIn('user_logins.country', ['India', 'US', 'UK'])
  })

onNull

Database
  .from('users')
  .join('user_logins', (query) => {
    query.onNull('user_logins.ip_address')
  })

onNotNull

Database
  .from('users')
  .join('user_logins', (query) => {
    query.onNotNull('user_logins.ip_address')
  })

onExists

Database
  .from('users')
  .join('user_logins', (query) => {
    query.onExists((subquery) => {
      subquery
        .select('*')
        .from('accounts')
        .whereRaw('users.account_id = accounts.id')
    })
  })

onNotExists

Database
  .from('users')
  .join('user_logins', (query) => {
    query.onNotExists((subquery) => {
      subquery
        .select('*')
        .from('accounts')
        .whereRaw('users.account_id = accounts.id')
    })
  })

onBetween

Database
  .from('users')
  .join('user_logins', (query) => {
    query.onBetween('user_logins.login_date', ['2020-10-01', '2020-12-31'])
  })

onNotBetween

Database
  .from('users')
  .join('user_logins', (query) => {
    query.onNotBetween('user_logins.login_date', ['2020-10-01', '2020-12-31'])
  })

having

The having method adds the having clause. It accepts the column name as the first argument, followed by the optional operator and the value.

Database
  .from('exams')
  .select('user_id')
  .groupBy('user_id')
  .having('score', '>', 80)

havingRaw

Most of the time, you will find yourself using the havingRaw method, as you can define the aggregates for the having clause.

Database
  .from('exams')
  .select('user_id')
  .groupBy('user_id')
  .havingRaw('SUM(score) > ?', [200])

having method variants

Following is the list of all the available having methods.

Method	Description
`havingIn`	Adds a having in clause to the query. It accepts an array of values.
`havingNotIn`	Adds a having not in clause to the query. It accepts an array of values.
`havingNull`	Adds a having null clause to the query.
`havingNotNull`	Adds a having not null clause to the query.
`havingExists`	Adds a having exists clause to the query.
`havingNotExists`	Adds a having not exists clause to the query.
`havingBetween`	Adds a having between clause to the query. It accepts an array of values.
`havingNotBetween`	Adds a having not between clause to the query. It accepts an array of values

distinct

The distinct method applies the distinct clause to the select statement. You can define one or more column names as multiple arguments.

Database
  .from('users')
  .distinct('country')

Database
  .from('users')
  .distinct('country', 'locale')

You can call the distinct method without any parameters to eliminate duplicate rows.

Database.from('users').distinct()

There is another PostgreSQL-only method, distinctOn. Here's an article explaining SELECT DISTINCT ON .

Database
  .from('logs')
  .distinctOn('url')
  .orderBy('created_at', 'DESC')

groupBy

The groupBy method applies the group by clause to the query.

Database
  .from('logs')
  .select('url')
  .groupBy('url')

groupByRaw

The groupByRaw method allows writing a SQL query to define the group by statement.

Database
  .from('sales')
  .select('year')
  .groupByRaw('year WITH ROLLUP')

orderBy

The orderBy method applies the order by clause to the query.

Database
  .from('users')
  .orderBy('created_at', 'desc')

You can sort by multiple columns by calling the orderBy method multiple times.

Database
  .from('users')
  .orderBy('username', 'asc')
  .orderBy('created_at', 'desc')

Or pass an array of objects.

Database
  .from('users')
  .orderBy([
    {
      column: 'username',
      order: 'asc',
    },
    {
      column: 'created_at',
      order: 'desc',
    }
  ])

Using sub queries

You can also pass a sub-query instance to the orderBy method — for example, Order posts by the number of comments they have received.

const commentsCountQuery = Database
  .from('comments')
  .count('*')
  .whereColumn('posts.id', '=', 'comments.post_id')

Database
  .from('posts')
  .orderBy(commentsCountQuery, 'desc')

orderByRaw

Use the orderByRaw method to define the sort order from a SQL string.

const commentsCountQuery = Database
  .raw(
    'select count(*) from comments where posts.id = comments.post_id'
  )
  .wrap('(', ')')

Database
  .from('posts')
  .orderBy(commentsCountQuery, 'desc')

offset

Apply offset to the SQL query

Database.from('posts').offset(11)

limit

Apply a limit to the SQL query

Database.from('posts').limit(20)

forPage

The forPage is a convenient method to apply offset and limit using the page number. It accepts a total of two arguments.

The first argument is the page number (not the offset).
The second argument is the number of rows to fetch. Defaults to 20

Database
  .from('posts')
  .forPage(request.input('page', 1), 20)

count

The count method allows you to use the count aggregate in your SQL queries.

The keys for the aggregate values are dialect-specific, and hence we recommend you always define aliases for predictable output.

In PostgreSQL, the count method returns a string representation of a bigint data type.

const users = await Database
  .from('users')
  .count('* as total')

console.log(users[0].total)

You can also define the aggregate as follows:

const users = await Database
  .from('users')
  .count('*', 'total')

console.log(users[0].total)

You can count multiple columns as follows:

const users = await Database
  .from('users')
  .count({
    'active': 'is_active',
    'total': '*',
  })

console.log(users[0].total)
console.log(users[0].active)

Other aggregate methods

The API for all the following aggregate methods is identical to the count method.

Method	Description
`countDistinct`	Count only the distinct rows
`min`	Aggregate values using the min function
`max`	Aggregate values using the max function
`sum`	Aggregate values using the sum function
`sumDistinct`	Aggregate values for only distinct rows using the sum function
`avg`	Aggregate values using the avg function
`avgDistinct`	Aggregate values for only distinct rows using the avg function

union

The union method allows to you build up a union query by using multiple instances of the query builder. For example:

Database
  .from('users')
  .whereNull('last_name')
  .union((query) => {
    query.from('users').whereNull('first_name')
  })

/**
SELECT * FROM "users" WHERE "last_name" IS NULL
UNION
SELECT * FROM "users" WHERE "first_name" IS NULL
*/

You can also wrap your union queries by passing a boolean flag as the 2nd argument.

Database
  .from('users')
  .whereNull('last_name')
  .union((query) => {
    query.from('users').whereNull('first_name')
  }, true) // 👈

/**
SELECT * FROM "users" WHERE "last_name" IS NULL
UNION
(SELECT * FROM "users" WHERE "first_name" IS NULL)
*/

You can pass an array of callbacks to define multiple union queries.

Database
  .from('users')
  .whereNull('last_name')
  .union([
    (query) => {
      query.from('users').whereNull('first_name')
    },
    (query) => {
      query.from('users').whereNull('email')
    },
  ], true)

/**
SELECT * FROM "users" WHERE "last_name" IS NULL
UNION
(SELECT * FROM "users" WHERE "first_name" IS NULL)
UNION
(SELECT * FROM "users" WHERE "email" IS NULL)
*/

Using sub queries

You can also define union queries by passing an instance of a query builder.

Database
  .from('users')
  .whereNull('last_name')
  .union([
    Database.from('users').whereNull('first_name'),
    Database.from('users').whereNull('email')
  ], true)

The following methods have the same API as the union method.

unionAll
intersect

with

The with method allows you to use CTE (Common table expression) in PostgreSQL, Oracle, SQLite3 and the MSSQL databases.

Database
  .query()
  .with('aliased_table', (query) => {
    query.from('users').select('*')
  })
  .select('*')
  .from('aliased_table')

/**
WITH "aliased_table" AS (
  SELECT * FROM "users"
)
SELECT * FROM "aliased_table"
*/

The method also accepts an optional third parameter which is an array of column names. The number of column names specified must match the number of columns in the result set of the CTE query.

Database
  .query()
  .with('aliased_table', (query) => {
    query.from('users').select('id', 'email')
  }, ['id', 'email'])
  .select('*')
  .from('aliased_table')

/**
WITH "aliased_table" (id, email) AS (
  SELECT * FROM "users"
)
SELECT * FROM "aliased_table"
*/

withMaterialized/withNotMaterialized

The withMaterialized and the withNotMaterialized methods allows you to use CTE (Common table expression) as materialized views in PostgreSQL and SQLite3 database.

Database
  .query()
  .withMaterialized('aliased_table', (query) => {
    query.from('users').select('*')
  })
  .select('*')
  .from('aliased_table')

/**
WITH "aliased_table" AS MATERIALIZED (
  SELECT * FROM "users"
)
SELECT * FROM "aliased_table"
*/

withRecursive

The withRecursive method creates a recursive CTE (Common table expression) in PostgreSQL, Oracle, SQLite3 and the MSSQL databases.

In the following example, we calculate the sum of all children accounts of a parent account. Also, we assume the following table structure.

id	name	parent_id	amount
1	Expenses	NULL	NULL
2	Car Expenses	1	100
3	Food Expenses	1	40
4	Earnings	NULL	NULL
5	Freelance work	4	100
6	Blog post payment	4	78
7	Car service	2	60

Database
  .query()
  .withRecursive('tree', (query) => {
    query
      .from('accounts')
      .select('amount', 'id')
      .where('id', 1)
      .union((subquery) => {
        subquery
          .from('accounts as a')
          .select('a.amount', 'a.id')
          .innerJoin('tree', 'tree.id', '=', 'a.parent_id')
      })
  })
  .sum('amount as total')
  .from('tree')

The above example is not meant to simplify the complexity of SQL. Instead, it demonstrates the power of the query builder to construct such SQL queries without writing them as a SQL string.

The method also accepts an optional third parameter which is an array of column names. The number of column names specified must match the number of columns in the result set of the CTE query.

Database
  .query()
  .withRecursive('tree', (query) => {
    query
      .from('accounts')
      .select('amount', 'id')
      .where('id', 1)
      .union((subquery) => {
        subquery
          .from('accounts as a')
          .select('a.amount', 'a.id')
          .innerJoin('tree', 'tree.id', '=', 'a.parent_id')
      })
  }, ['amount', 'id'])
  .sum('amount as total')
  .from('tree')

Here's a great article explaining the PostgreSQL Recursive Query

update

The update method allows updating one or more database rows. You can make use of the query builder to add additional constraints when performing the update.

const affectedRows = Database
  .from('users')
  .where('id', 1)
  .update({ email: 'virk@adonisjs.com' })

The return value is the number of affected rows. However, when using PostgreSQL, Orcale, or MSSQL, you can specify the return columns as well.

const rows = Database
  .from('users')
  .where('id', 1)
  .update(
    { email: 'virk@adonisjs.com' },
    ['id', 'email'] // columns to return
  )

console.log(rows[0].id)
console.log(rows[0].email)

increment

The increment method allows incrementing the value for one or more columns.

Database
  .from('accounts')
  .where('id', 1)
  .increment('balance', 10)

/**
UPDATE "accounts"
SET
  "balance" = "balance" + 10
WHERE
  "id" = 1
*/

You can also increment multiple columns by passing an object.

Database
  .from('accounts')
  .where('id', 1)
  .increment({
    balance: 10,
    credit_limit: 5
  })

/**
UPDATE "accounts"
SET
  "balance" = "balance" + 10,
  "credit_limit" = "credit_limit" + 5
WHERE
  "id" = 1
*/

decrement

The decrement method is the opposite of the increment method. However, the API is the same.

Database
  .from('accounts')
  .where('id', 1)
  .decrement('balance', 10)

delete

The delete method issues a delete SQL query. You can make use of the query builder to add additional constraints when performing the delete.

Database
  .from('users')
  .where('id', 1)
  .delete()

The delete method also has an alias called del.

useTransaction

The useTransaction method instructs the query builder to wrap the query inside a transaction. The guide on database transactions covers different ways to create and use transactions in your application.

const trx = await Database.transaction()

Database
  .from('users')
  .useTransaction(trx) // 👈
  .where('id', 1)
  .update({ email: 'virk@adonisjs.com' })

await trx.commit()

forUpdate

The forUpdate method acquires an update lock on the selected rows in PostgreSQL and MySQL.

Make sure always to supply the transaction object using the useTransaction method before using forUpdate or similar locks.

const user = Database
  .from('users')
  .where('id', 1)
  .useTransaction(trx)
  .forUpdate() // 👈
  .first()

forShare

The forShare adds a FOR SHARE in PostgreSQL and a LOCK IN SHARE MODE for MySQL during a select statement.

const user = Database
  .from('users')
  .where('id', 1)
  .useTransaction(trx)
  .forShare() // 👈
  .first()

skipLocked

The skipLocked method skips the rows locked by another transaction. The method is only supported by MySQL 8.0+ and PostgreSQL 9.5+.

Database
  .from('users')
  .where('id', 1)
  .forUpdate()
  .skipLocked() // 👈
  .first()

/**
SELECT * FROM "users"
WHERE "id" = 1
FOR UPDATE SKIP LOCKED
*/

noWait

The noWait method fails if any of the selected rows are locked by another transaction. The method is only supported by MySQL 8.0+ and PostgreSQL 9.5+.

Database
  .from('users')
  .where('id', 1)
  .forUpdate()
  .noWait() // 👈
  .first()

/**
SELECT * FROM "users"
WHERE "id" = 1
FOR UPDATE NOWAIT
*/

clone

The clone method returns a new query builder object with all constraints applied from the original query.

const query = Database.from('users').select('id', 'email')
const clonedQuery = query.clone().clearSelect()

await query // select "id", "email" from "users"
await clonedQuery // select * from "users"

debug

The debug method allows enabling or disabling debugging at an individual query level. Here's a complete guide on debugging queries.

Database
  .from('users')
  .debug(true)

timeout

Define the timeout for the query. An exception is raised after the timeout has been exceeded.

The value of timeout is always in milliseconds.

Database
  .from('users')
  .timeout(2000)

You can also cancel the query when using timeouts with MySQL and PostgreSQL.

Database
  .from('users')
  .timeout(2000, { cancel: true })

toSQL

The toSQL method returns the query SQL and bindings as an object.

const output = Database
  .from('users')
  .where('id', 1)
  .toSQL()

console.log(output)

The toSQL object also has the toNative method to format the SQL query as per the database dialect in use.

const output = Database
  .from('users')
  .where('id', 1)
  .toSQL()
  .toNative()

console.log(output)

toQuery

Returns the SQL query after applying the bind params.

const output = Database
  .from('users')
  .where('id', 1)
  .toQuery()

console.log(output)
// select * from "users" where "id" = 1

Executing queries

The query builder extends the native Promise class. You can execute the queries using the await keyword or chaining the then/catch methods.

Database
  .from('users')
  .then((users) => {
  })
  .catch((error) => {
  })

Using async/await

const users = await Database.from('users')

Also, you can execute a query by explicitly calling the exec method.

const users = await Database.from('users').exec()

first

The select queries always return an array of objects, even when the query is intended to fetch a single row. However, using the first method will give you the first row or null (when there are no rows).

First does NOT mean the first row in the table. It means the first row from the results array in whatever order you fetched it from the database.

const user = await Database
  .from('users')
  .where('id', 1)
  .first()

if (user) {
  console.log(user.id)
}

firstOrFail

The firstOrFail method is similar to the first method except, it raises an exception when no rows are found.

const user = await Database
  .from('users')
  .where('id', 1)
  .firstOrFail()

Pagination

The query builder has first-class support for offset-based pagination. It also automatically counts the number of total rows by running a separate query behind the scenes.

const page = request.input('page', 1)
const limit = 20

const results = await Database
  .from('users')
  .paginate(page, limit)

The paginate method returns an instance of the SimplePaginator class. The class has the following properties and methods.

firstPage

Returns the number for the first page. It is always 1.

results.firstPage

perPage

Returns the value for the limit passed to the paginate method.

results.perPage

currentPage

Returns the value of the current page.

results.currentPage

lastPage

Returns the value for the last page by taking the total of rows into account.

results.lastPage

total

Holds the value for the total number of rows in the database.

results.total

hasPages

A boolean to know if there are pages for pagination. You can rely on this value to decide when or when not to show the pagination links.

Following is an example of the Edge view.

@if(results.hasPages)

  {{-- Display pagination links --}}

@endif

hasMorePages

A boolean to know if there are more pages to go after the current page.

results.hasMorePages

all()

Returns an array of rows returned by the SQL queries.

results.all()

getUrl

Returns the URL for a given page number.

result.getUrl(1) // /?page=1

getNextPageUrl

Returns the URL for the next page

// Assuming the current page is 2

result.getNextPageUrl() // /?page=3

getPreviousPageUrl

Returns the URL for the previous page

// Assuming the current page is 2

result.getPreviousPageUrl() // /?page=1

getUrlsForRange

Returns URLs for a given range. Helpful when you want to render links for a given range.

Following is an example of using getUrlsForRange inside an Edge template.

@each(
  link in results.getUrlsForRange(results.firstPage, results.lastPage)
)
  <a
    href="{{ link.url }}"
    class="{{ link.isActive ? 'active' : '' }}"
  >
    {{ link.page }}
  </a>
@endeach

toJSON

The toJSON method returns an object with meta and data properties. The output of the method is suitable for JSON API responses.

results.toJSON()

/**
{
  meta: {
    total: 200,
    per_page: 20,
    current_page: 1,
    first_page: 1,
    last_page: 20,
    ...
  },
  data: [
    {
    }
  ]
}
*/

baseUrl

All of the URLs generated by the paginator class use the / (root) URL. However, you can change this by defining a custom base URL.

results.baseUrl('/posts')

results.getUrl(2) // /posts?page=2

queryString

Define query string to be appended to the URLs generated by the paginator class.

results.queryString({ limit: 20, sort: 'top' })

results.getUrl(2) // /?page=2&limit=20&sort=top

Helpful properties and methods

Following is the list of properties and methods you may occasionally need when building something on top of the query builder.

client

Reference to the instance of the underlying database query client .

const query = Database.query()
console.log(query.client)

knexQuery

Reference to the instance of the underlying KnexJS query.

const query = Database.query()
console.log(query.knexQuery)

hasAggregates

A boolean to know if the query is using any of the aggregate methods.

const query = Database.from('posts').count('* as total')
console.log(query.hasAggregates) // true

hasGroupBy

A boolean to know if the query is using a group by clause.

const query = Database.from('posts').groupBy('tenant_id')
console.log(query.hasGroupBy) // true

hasUnion

A boolean to know if the query is using a union.

const query = Database
  .from('users')
  .whereNull('last_name')
  .union((query) => {
    query.from('users').whereNull('first_name')
  })

console.log(query.hasUnion) // true

clearSelect

Call this method to clear selected columns.

const query = Database.query().select('id', 'title')
query.clone().clearSelect()

clearWhere

Call this method to clear where clauses.

const query = Database.query().where('id', 1)
query.clone().clearWhere()

clearOrder

Call this method to clear the order by constraint.

const query = Database.query().orderBy('id', 'desc')
query.clone().clearOrder()

clearHaving

Call this method to clear the having clause.

const query = Database.query().having('total', '>', 100)
query.clone().clearHaving()

clearLimit

Call this method to clear the applied limit.

const query = Database.query().limit(20)
query.clone().clearLimit()

clearOffset

Call this method to clear the applied offset.

const query = Database.query().offset(20)
query.clone().clearOffset()

reporterData

The query builder emits the db:query event and reports the query's execution time with the framework profiler.

Using the reporterData method, you can pass additional details to the event and the profiler.

const query = Database.from('users')

await query
  .reporterData({ userId: auth.user.id })
  .select('*')

Within the db:query event, you can access the value of userId as follows.

Event.on('db:query', (query) => {
  console.log(query.userId)
})

withSchema

Specify the PostgreSQL schema to use when executing the query.

Database
  .from('users')
  .withSchema('public')
  .select('*')

as

Specify the alias for a given query. Usually helpful when passing the query builder instance as a subquery. For example:

Database
  .from('users')
  .select(
    Database
      .from('user_logins')
      .select('ip_address')
      .whereColumn('users.id', 'user_logins.user_id')
      .orderBy('id', 'desc')
      .limit(1)
      .as('last_login_ip') // 👈 Query alias
  )

if

The if helper allows you to conditionally add constraints to the query builder. For example:

Database
  .from('users')
  .if(searchQuery, (query) => {
    query.where('first_name', 'like', `%${searchQuery}%`)
    query.where('last_name', 'like', `%${searchQuery}%`)
  })

You can define the else method by passing another callback as the second argument.

Database
  .from('users')
  .if(
    condition,
    (query) => {}, // if condition met
    (query) => {}, // otherwise execute this
  )

unless

The unless method is opposite of the if helper.

Database
  .from('projects')
  .unless(filters.status, () => {
    /**
     * Fetch projects with "active" status when
     * not status is defined in filters
     */
    query.where('status', 'active')
  })

You can pass another callback which gets executed when the unless statement isn't true.

Database
  .from('users')
  .unless(
    condition,
    (query) => {}, // if condition met
    (query) => {}, // otherwise execute this
  )

match

The match helper allows you define an array of conditional blocks to match from and execute the corresponding callback.

In the following example, the query builder will go through all the conditional blocks and executes the first matching one and discards the other. Think of it as a switch statement in JavaScript.

Database
  .query()
  .match(
    [
      // Run this is user is a super user
      auth.isSuperUser, (query) => query.whereIn('status', ['published', 'draft'])
    ],
    [
      // Run this is user is loggedin
      auth.user, (query) => query.where('user_id', auth.user.id)
    ],
    // Otherwise run this
    (query) => query.where('status', 'published').where('is_public', true)
  )