v8.0.0

This isn't a breaking change that will affect most people. In fact, it will most likely improve your code.

qb.from( "users" )
    .where( "active", 1 )
    .when( len( url.q ), function( q ) {
        q.where( "username", "LIKE", q & "%" )
            .orWhere( "email", "LIKE", q & "%" );   
    } );

Would generate the following SQL:

SELECT *
FROM "users"
WHERE "active" = ?
    AND "username" = ?
    OR "email" = ?

The problem with this statement is that the OR can short circuit the active check.

The fix is to wrap the LIKE statements in parenthesis. This is done in qb using a function callback to where.

qb.from( "users" )
    .where( "active", 1 )
    .where( function( q ) {
        q.where( "username", "LIKE", q & "%" )
            .orWhere( "email", "LIKE", q & "%" );
    } );

SELECT *
FROM "users"
WHERE "active" = ?
    AND (
        "username" = ?
        OR "email" = ?
    )

When using the when control flow function, it was easy to miss this. This is because you are already in a closure - it looks the same as when using where to group the clauses.

In qb 8.0.0, when will automatically group added where clauses when needed. That means our original example now produces the SQL we probably expected.

// qb 8.0.0
qb.from( "users" )
    .where( "active", 1 )
    .when( len( url.q ), function( q ) {
        q.where( "username", "LIKE", q & "%" )
            .orWhere( "email", "LIKE", q & "%" );   
    } );

SELECT *
FROM "users"
WHERE "active" = ?
    AND (
        "username" = ?
        OR "email" = ?
    )

Grouping is not needed if there is no OR combinator. In these cases no grouping is added.

// qb 8.0.0
qb.from( "users" )
    .where( "active", 1 )
    .when( url.keyExists( "admin" ), function( q ) {
        q.where( "admin", 1 )
            .whereNotNull( "hireDate" );
    } );

SELECT *
FROM "users"
WHERE "active" = ?
    AND "admin" = ?
    AND "hireDate IS NOT NULL

If you had already wrapped your expression in a group inside the when callback, nothing changes. Your code works as before. The OR combinator check only works on the top most level of added where clauses.

qb.from( "users" )
    .where( "active", 1 )
    .when( len( url.q ), function( q ) {
        q.where( function( q2 ) {
            q2.where( "username", "LIKE", q & "%" )
                .orWhere( "email", "LIKE", q & "%" );
        } );
    } );

SELECT *
FROM "users"
WHERE "active" = ?
    AND (
        "username" = ?
        OR "email" = ?
    )

Additionally, if you do not add any where clauses inside a when callback, nothing changes from qb 7.

The breaking change part is if you were relying on these statements residing at the same level without grouping. In those cases, you may pass the withoutScoping flag to the when callback.

// qb 8.0.0
qb.from( "users" )
    .where( "active", 1 )
    .when(
        condition = len( url.q ),
        onTrue = function( q ) {
            q.where( "username", "LIKE", q & "%" )
                .orWhere( "email", "LIKE", q & "%" );   
        },
        withoutScoping = true
    );

SELECT *
FROM "users"
WHERE "active" = ?
    AND "username" = ?
    OR "email" = ?

v7.0.0

Lucee 4.5 and Adobe ColdFusion 11 EOL

Support for Lucee 4.5 and Adobe ColdFusion 11 has been dropped. If you need support for these engines, please remain on an earlier version of qb.

MSSQLGrammar renamed to SqlServerGrammar

MSSQLGrammar was visually too close to MySQLGrammar and was hard to differentiate quickly. SqlServerGrammar is much more unique and easily identifiable. Additionally, more people that use this library refer to their database engine as "SQL Server" than "MSSQL".

Variadic Parameters Support Removed

Variadic parameter support was the ability to pass any number of arguments to certain methods like select.

qb.select( "name", "email", "createdDate" );

This code came with a slight performance cost and readability cost. That, combined with the fact that the above syntax is very close to an array, we are dropping support for variadic parameters. To migrate, wrap instances of variadic parameters in an array:

qb.select( [ "name", "email", "createdDate" ] );

defaultGrammar updated to be the full WireBox mapping

In previous versions, the value passed to defaultGrammar was used to look up a mapping in the @qb namespace. This made it difficult to add or use grammars that weren't part of qb. (You could get around this be registering your custom grammar in the @qb namespace, but doing so seemed strange.)

To migrate this code, change your defaultGrammar to be the full WireBox mapping in your moduleSettings:

moduleSettings = {
    "qb": {
        "defaultGrammar": "MSSQLGrammar@qb"
    }
};

value method argument order changed

A defaultValue parameter and optional exception throwing was added to value. This pushed the options struct to the end of the method. If you are using positional parameters with value, you will need to update your method calls to either use named parameters or the new positions.

public any function value(
    required string column,
    string defaultValue = "",
    boolean throwWhenNotFound = false,
    struct options = {}
);

Some methods renamed callback to query

All methods that could conceivably take a subquery as well as a value now accept a closure or another builder instance to use as a subquery. This led to changing the callback argument to query in the following cases:

• whereSub

• whereInSub

• whereExists

• orWhereExists

• whereNotExists

• andWhereNotExists

• orWhereNotExists

• whereNullSub

• orderBySub

• subSelect

If you are using named parameters with any of the above methods you will need to migrate your method calls.

v5.0.0

Version v5.0.0 brings support for SchemaBuilder inside qb. To avoid naming confusion, Builder was renamed to QueryBuilder. Any references in your code to Builder@qb need to be updated to QueryBuilder@qb.

Installation

Usage

To start a new query, instantiate a new Builder: wirebox.getInstance('QueryBuilder@qb').

By default, qb uses a generic Grammar. You can specify your specific grammar in ColdBox by setting the defaultGrammar in your moduleSettings.

moduleSettings = {
    qb = {
        defaultGrammar = "MySQLGrammar@qb"
    }
};

The grammars provided by qb are:

• MySQLGrammar

• OracleGrammar

• PostgresGrammar

• SqlServerGrammar

If you are not using WireBox, just make sure to wire up the Builder object with the correct grammar:

var grammar = new qb.models.Grammars.MySQLGrammar();
var builder = new qb.models.Query.QueryBuilder( grammar );

SQL Type Inferral

QB binds all parameters by default and guesses the SQL type based on passed values. The default SQL type for numeric values is CF_SQL_NUMERIC, which is a floating point number, for the widest compatibility. This can cause performance problems with large recordsets in some database engines. You can provide a different default in coldbox.cfc if you wish to override this setting:

moduleSettings = {
    qb = {
        defaultGrammar = "MySQLGrammar@qb",
        numericSQLType = "CF_SQL_BIGINT"
    }
};

Integrating With FW/1

Note: These instructions assume a basic knowledge of FW/1, a working FW/1 application structure with qb installed in the /subsystems directory (manually or via CommandBox), and a database configured to run with your application.

Wiring Up With DI/1

Once the application structure is setup, now we need to wire up qb to a bean factory using DI/1.

First we will add a mapping in Application.cfc.

this.mappings = {
    "/qb" = expandPath("./subsystems/qb")
};

Next we need to tell DI/1 where qb's components are and how to reference them for later use in the application. We can do so by defining the configuration settings in the variables.framework.subsystems struct in Application.cfc. The example below makes use of a load listener to declare each component instance and pass in any constructor arguments.

qb = {
  diLocations = "/qb/models",
  diConfig = {
    loadListener = function( di1 ) {
      di1.declare( "BaseGrammar" ).instanceOf( "qb.models.Query.Grammars.Grammar" ).done()
         .declare( "MySQLGrammar" ).instanceOf( "qb.models.Query.Grammars.MySQLGrammar" ).done()
         .declare( "QueryUtils" ).instanceOf( "qb.models.Query.QueryUtils" ).done()
         .declare( "QueryBuilder" ).instanceOf( "qb.models.Query.QueryBuilder" )
         .withOverrides({
            grammar = di1.getBean( "MySQLGrammar" ),
            utils = di1.getBean( "QueryUtils" ),
            returnFormat = "array"
         })
         .asTransient();
    }
  }
}

Usage In Your FW/1 Application

Now that everything is configured, you can launch your application with CommandBox by entering start in the terminal or use whatever method you're accustomed to.

To access qb from your application's code, you can call on it by using getBeanFactory().

// Create an instance of qb
builder = getBeanFactory( "qb" ).getBean( "QueryBuilder" );
// Query the database
posts = builder.from( "Posts" ).get();
posts = builder.from( "Posts" ).where( "IsDraft", "=", 0 ).get();

A query builder is a stateful, transient object. That means that if you want to execute two different queries, you need two separate instances of QueryBuilder.

As such, be careful when injecting QueryBuilder in to a component. If the component is a singleton, you will need to create the QueryBuilder inline or use a provider. This applies to ColdBox handlers as well.

While the above may seem innoculous, it can run in to issues as multiple requests come in to your application. Each request is sharing the same query builder instance and subsequent requests will have unintended results as the where clause keeps growing request after request.

The solution is to either create the QueryBuilder inline, ensuring that each request has its own query to execute:

Or to use a WireBox provider to create a new query each time it is accessed:

newQuery

Once you have access to a QueryBuilder instance, you can create a new query using the same datasource, utils, returnFormat, paginationCollector, columnFormatter, and defaultOptions as the current QueryBuilder instance.

We welcome all types of contributions!

The most common type of contribution is to fix an incorrect SQL generation for a database grammar.

To debug what SQL is being ran, you can always call toSQL on any QueryBuilder or SchemaBuilder object. Additionally, you can listen to the preQBExecute interception point for the generated SQL.

Each of the database grammars have two tests — {Grammar}QueryBuilderSpec.cfc and {Grammar}SchemaBuilderSpec.cfc. These tests run the same qb syntax across the different grammars. In each test are methods that return SQL strings like so:

// MSSQLQueryBuilderSpec.cfc
function orWhere() {
    // If just a string is returned, we assume the bindings is an empty array ([])
    return {
        sql = "SELECT * FROM [users] WHERE [id] = ? OR [email] = ?",
        bindings = [ 1, "foo" ]
    };
}

// OracleSchemaBuilderSpec.cfc
function boolean() {
    // returns an array since schema builder can execute multiple statements.
    return [ "CREATE TABLE ""USERS"" (""ACTIVE"" NUMBER(1, 0) NOT NULL)" ];
}

If you find an issue with the SQL generated from a grammar, please file a pull request with the correct SQL in these tests. It's okay if you don't submit a fix as well. (But we'd greatly appreciate it!) Doing so will help expedite the fix.

If you want to add support for a new database grammar, simply copy these two tests from an existing grammar, rename them, change the getBuilder method to return your new grammar, and fill out the SQL as it should be. That will guide your implementation to be 100% compatible with the other grammars in qb.

8.6.1

• Correctly wrap CTE expressions with parenthesis when required in certain grammars.

8.6.0

8.5.0

QueryBuilder

• Correct return aggregate values for date values from max and min executors.

• Correctly format a COUNT(DISTINCT column) query.

• Only use bulk insert syntax when needed in OracleGrammar due to interactions between the result parameter to cfquery, Lucee, and the Oracle JDBC driver.

SchemaBuilder

8.4.9

• Swap master branch to main branch.

8.4.8

• Remove unnecessary injection for QueryUtils.

8.4.7

• Account for raw expressions when generating mementos for comparison

8.4.6

8.4.5

8.4.1 - 8.4.4

• Migrate release process to GitHub Actions.

8.4.0

• Add a simplePaginate pagination method for quicker performance when total records or total pages are not needed or too slow.

8.3.0

8.2.2

• Default to html for the dump format argument to writeDump.

8.2.1

• Correctly use the passed in strictDateDetection to the QueryUtils.cfc.

8.2.0

8.1.0

8.0.3

• Ignore select bindings for aggregate queries.

• Allow spaces in table aliases.

• Split FLOAT and DECIMAL column types in SQL Server.

8.0.2

• Clear orderBy bindings when calling clearOrders.

8.0.1

• Trim table definitions before searching for aliases. Makes qb more lenient with extra whitespace.

8.0.0

BREAKING CHANGES

Other Changes

7.10.0

• Expose nested where functions to enable advanced query manipulation in downstream libraries like Quick.

7.9.9

• Fixes for OracleGrammar including table aliases and wrapped subqueries.

7.9.8

7.9.7

7.9.6

7.9.5

• Handle enhanced numeric checks with Secure Profile enabled.

7.9.4

• Allow raw statements in basic where clauses.

7.9.3

7.9.2

7.9.1

• Handle multi-word columns in queryRemoveColumns.

7.9.0

• Remove elvis operator due to ACF compatibility issues

7.8.0

7.7.3

7.7.2

• Compatibility fix for ACF 2018 and listLast parsing.

• Ignore table qualifiers for insert and update.

7.7.1

• Fix a bug with preventDuplicateJoins when using the closure syntax with a join.

7.7.0

• Add executionTime to the data output from BaseGrammar, including being available in interceptors.

7.6.2

• Fix a case where a column was not wrapped correctly when a where used a subquery for the value.

7.6.1

• Avoid duplicate function due to cbORM / Hibernate bugs when used in the same application.

7.6.0

• Split off a private whereBasic method. This is used in Quick to provide extra sql type features.

7.5.1

Fixed an issue using column formatters with update and insert.

7.5.0

Using a new preventDuplicateJoins setting in the module settings, qb can detect duplicate joins and ignore them. This is especially useful in a heavily filtered and dynamic query where you may or may not need the join at all or more than one column may need the same join. preventDuplicateJoins defaults to false, so it is opt-in. It may be turned on by default in a future breaking release of qb.

7.4.0

You can now use two shortcut methods: orderByAsc and orderByDesc. Additionally, orderBySub or using orderBy with a closure or builder instance will respect the direction argument.

7.3.15

7.3.14

7.3.13

7.3.12

7.3.9, 7.3.10, 7.3.11

7.3.8

7.3.7

7.3.5, 7.3.6

7.3.4

7.3.2, 7.3.3

7.3.1

• Fix for null values breaking the new checkIsActuallyNumeric method in QueryUtils.

7.3.0

• Add a parameterLimit public property to SqlServerGrammar. This property is used in Quick to split up eager loading to work around the 2100 param limit of SQL Server.

7.2.0

7.1.0

• Lambdas (arrow functions) are now allowed wherever closures are allowed.

7.0.0

BREAKING CHANGES

• Drop support for Lucee 4.5 and Adobe ColdFusion 11.

• MSSQLGrammar renamed to SqlServerGrammar

• Remove variadic parameters support in builder functions like select.

• The defaultGrammar mapping needs to be the full WireBox mapping, including the @qb, if needed.

  • For instance, MSSQLGrammar would become MSSQLGrammar@qb.

  • This will allow for other grammars to be more easily contributed via third party modules.

• The argument names of forPage changed to match the new paginate method.

• Add defaultValue and optional exception throwing to value. (This changed the argument order.)

• All methods that could conceivably take a subquery as well as a value now accept a closure or another builder instance to use as a subquery. (This changed the argument names in some instances.)

Other Changes

• Completely revamped documentation! (You're looking at it right now.)

• Preserve column case and order when converting a query to an array using the default "array" return format.

• Allow closures to be used in left and right joins.

• Add raw in alterTable segments.

• Add dropAllObjects support for SqlServerGrammar and OracleGrammar to support migrate fresh from cfmigrations.

• Add a renameTable alias for rename.

• Remove default constraints when dropping columns with a default on SqlServerGrammar.

• Add more column types and column helpers to SchemaBuilder, including:

  • datetimeTz

  • lineString

  • nullableTimestamps

  • point

  • polygon

  • softDeletes

  • softDeletesTz

  • timeTz

  • timestamps

  • timestampTz

  • timestampsTz

  • withCurrent

6.4.0

• Allow Expressions (query.raw) in update statements.

Introduction

Using qb, you can:

• Quickly scaffold simple queries

• Make complex, out-of-order queries possible

• Abstract away differences between database engines

Requirements

• Adobe ColdFusion 2016+

• Lucee 5+

qb supports four major database grammars:

• MySQL (MySQLGrammar)

• Oracle (OracleGrammar)

• Postgres (PostgresGrammar)

• Microsoft SQL Server (SqlServerGrammar)

Discussion & Help

The Box modules discussion group and community can be found here:

Installation

Code Samples

Compare these two examples:

// Plain old CFML
q = queryExecute("SELECT * FROM users");

// qb
query = wirebox.getInstance('QueryBuilder@qb');
q = query.from('users').get();

The differences become even more stark when we introduce more complexity:

// Plain old CFML
q = queryExecute(
    "SELECT * FROM posts WHERE published_at IS NOT NULL AND author_id IN ?",
    [ { value = '5,10,27', cfsqltype = 'CF_SQL_NUMERIC', list = true } ]
);

// qb
query = wirebox.getInstance('QueryBuilder@qb');
q = query.from('posts')
         .whereNotNull('published_at')
         .whereIn('author_id', [5, 10, 27])
         .get();

With qb you can easily handle setting order by statements before the columns you want or join statements after a where clause:

query = wirebox.getInstance('QueryBuilder@qb');
q = query.from('posts')
         .orderBy('published_at')
         .select('post_id', 'author_id', 'title', 'body')
         .whereLike('author', 'Ja%')
         .join('authors', 'authors.id', '=', 'posts.author_id')
         .get();

// Becomes

q = queryExecute(
    "SELECT post_id, author_id, title, body FROM posts INNER JOIN authors ON authors.id = posts.author_id WHERE author LIKE ? ORDER BY published_at",
    [ { value = 'Ja%', cfsqltype = 'CF_SQL_VARCHAR', list = false, null = false } ]
);

qb enables you to explore new ways of organizing your code by letting you pass around a query builder object that will compile down to the right SQL without you having to keep track of the order, whitespace, or other SQL gotchas!

Usage

To start a new query, instantiate a new Builder: wirebox.getInstance('QueryBuilder@qb').

By default, qb uses a generic Grammar. You can specify your specific grammar in ColdBox by setting the defaultGrammar in your moduleSettings.

moduleSettings = {
    qb = {
        defaultGrammar = "MySQLGrammar@qb"
    }
};

If you are not using WireBox, just make sure to wire up the Builder object with the correct grammar:

var grammar = new qb.models.Query.Grammars.MySQLGrammar();
var builder = new qb.models.Query.Builder( grammar );

Join clauses range from simple to complex including joining complete subqueries on multiple conditions. qb has your back with all of these use cases.

join

Applies a join to the query. The simplest join is to a table based on two columns:

When doing a simple join using = as the operator, you can omit it and pass just the column names:

When you need to specify more clauses to join, you can pass a function as the second argument:

Conditions inside a join clause can be grouped using a function.

joinWhere

Adds a join to another table based on a WHERE clause instead of an ON clause. WHERE clauses introduce parameters and parameter bindings whereas on clauses join between columns and don't need parameter bindings.

For simple joins, this specifies a column on which to join the two tables:

joinRaw

joinSub

Alternatively, a function may be used to define the derived table:

Complex join conditions are also possible by passing a function as the third parameter:

leftJoin

leftJoinRaw

leftJoinSub

rightJoin

rightJoinRaw

rightJoinSub

crossJoin

crossJoinRaw

Uses the raw SQL provided to as the table for the cross join clause. Cross joins cannot be further constrained with on or where clauses.

crossJoinSub

newJoin

JoinClause

on

Applies a join condition to the JoinClause. An alias for whereColumn.

orOn

Applies a join condition to the JoinClause using an or combinator. An alias for orWhereColumn.

Preventing Duplicate Joins

You can optionally configure qb to ignore duplicate joins. With this setting turned on each JoinClause is inspected and checked if it matches any existing JoinClause instances on the query. This is useful if you have a table shared between optional constraints and want to ensure it is only added once.

You can opt-in to this behavior by setting preventDuplicateJoins = true in your moduleSettings in config/ColdBox.cfc.

Common Table Expressions (CTEs) are powerful SQL concept that allow you to create re-usable temporal result sets, which can be referenced as a table within your SQL. CTEs are available in many common database engines and are available in latest versions of all of the support grammars.

CTEs come in two basic types:

• Non-recursive — These are statements that do not reference themselves, in simplified terms they are like a derived table that can be referenced by a user-defined name.

• Recursive — Recursive CTEs reference themselves and are generally used for creating hierarchical data—such as creating a parent/child relationship within a table.

While all of the grammars currently support CTEs, there is enough difference between the various databases implementations of CTEs that unless your CTEs are fairly basic, using CTEs within your project will most likely tie your project to a specific database, unless you account for the differences in your code.

However, CTEs are can be extremely useful to solve certain use cases.

To add CTEs to your queries, you have two methods available:

• with() — Allows you to define a non-recursive CTE.

• withRecursive() — Allows you to define a recursive CTE.

with

You can build a CTE using a function:

// qb
query.with( "UserCTE", function ( q ) {
        q
            .select( [ "fName as firstName", "lName as lastName" ] )
            .from( "users" )
            .where( "disabled", 0 );
    } )
    .from( "UserCTE" )
    .get();

WITH `UserCTE` AS (
    SELECT
        `fName` as `firstName`,
        `lName` as `lastName`
    FROM `users`
    WHERE `disabled` = 0
) SELECT * FROM `UserCTE`

Alternatively, you can use a QueryBuilder instance instead of a function:

// qb
var cte = query
    .select( [ "fName as firstName", "lName as lastName" ] )
    .from( "users" )
    .where( "disabled", 0 );

query.with( "UserCTE", cte )
    .from( "UserCTE" )
    .get();

WITH `UserCTE` AS (
    SELECT
        `fName` as `firstName`,
        `lName` as `lastName`
    FROM `users`
    WHERE `disabled` = 0
)
SELECT * FROM `UserCTE`

A single query can reference multiple CTEs:

query.with( "UserCTE", function ( q ) {
        q.select( [ "id", "fName as firstName", "lName as lastName" ] )
            .from( "users" )
            .where( "disabled", 0 );
    } )
    .with( "BlogCTE", function ( q ) {
        q.from( "blogs" )
            .where( "disabled", 0 );
    } )
    .from( "BlogCTE as b" )
    .join( "UserCTE as u", "b.Creator", "u.id" )
    .get();

WITH `UserCTE` AS (
    SELECT
        `id`,
        `fName` as `firstName`,
        `lName` as `lastName`
    FROM `users`
    WHERE `disabled` = 0
),
`BlogCTE` AS (
    SELECT *
    FROM `blogs`
    WHERE `disabled` = 0
)
SELECT *
FROM `BlogCTE` AS `b`
INNER JOIN `UserCTE` AS `u`
ON `b`.`Creator` = `u`.`id`

withRecursive

Here is an example of building a recursive CTE using SQL Server which would return all parent/child rows and show their generation/level depth:

query
.withRecursive( "Hierarchy", function ( q ) {
    q.select( [ "Id", "ParentId", "Name", q.raw( "0 AS [Generation]" ) ] )
        .from( "Sample" )
        .whereNull( "ParentId" )
        // use recursion to join the child rows to their parents
        .unionAll( function ( q ) {
            q.select( [
                    "child.Id",
                    "child.ParentId",
                    "child.Name",
                    q.raw( "[parent].[Generation] + 1" )
                ] )
                .from( "Sample as child" )
                .join( "Hierarchy as parent", "child.ParentId", "parent.Id" );
        } );
    }, [ "Id", "ParentId", "Name", "Generation" ] )
    .from( "Hierarchy" )
    .get();

WITH [Hierarchy] ([Id], [ParentId], [Name], [Generation]) AS (
    SELECT
        [Id],
        [ParentId],
        [Name],
        0 AS [Generation]
    FROM [Sample]
    WHERE [ParentId] IS NULL
    UNION ALL
    SELECT
        [child].[Id],
        [child].[ParentId],
        [child].[Name],

[parent].[Generation] + 1
    FROM [Sample] AS [child]
    INNER JOIN [Hierarchy] AS [parent]
        ON [child].[ParentId] = [parent].[Id]
) SELECT * FROM [Hierarchy]

The orderBy method seems simple but has a lot of depth depending on the type of arguments you pass in.

Order By (String)

query.from( "users" )
    .orderBy( "email" );

SELECT *
FROM `users`
ORDER BY `email` ASC

Calling orderBy multiple times will append to the order list.

query.from( "users" )
    .orderBy( "email" )
    .orderBy( "username", "desc" );

SELECT *
FROM `users`
ORDER BY
  `email` ASC,
  `username` DESC

query.from( "users" )
    .orderBy( query.raw( "DATE(created_at)" ) );

SELECT *
FROM `users`
ORDER BY DATE(created_at)

Order By (List)

query.from( "users" )
    .orderBy( "email|asc,username", "desc" );

SELECT *
FROM `users`
ORDER BY
  `email` ASC,
  `username` DESC

Order By (Array of Strings)

query.from( "users" )
    .orderBy( [ "email|asc", "username" ], "desc" );

SELECT *
FROM `users`
ORDER BY
  `email` ASC,
  `username` DESC

Order By (Array of Structs)

query.from( "users" )
    .orderBy( [
        { "column": "email", "direction": "asc" },
        "username"
    ], "desc" );

SELECT *
FROM `users`
ORDER BY
  `email` ASC,
  `username` DESC

Order By (Subquery)

You can order with a subquery using either a function or a QueryBuilder instance.

query.from( "users" )
    .orderBy( function( q ) {
        q.selectRaw( "MAX(created_date)" )
            .from( "logins" )
            .whereColumn( "users.id", "logins.user_id" );
    } );

SELECT *
FROM `users`
ORDER BY (
    SELECT MAX(created_date)
    FROM `logins`
    WHERE `users`.`id` = `logins`.`user_id`
)

Order By Raw

query.from( "users" )
    .orderByRaw( "CASE WHEN status = ? THEN 1 ELSE 0 END DESC", [ 1 ] );

SELECT *
FROM `users`
ORDER BY CASE WHEN status = ? THEN 1 ELSE 0 END DESC

clearOrders

query.from( "users" )
    .orderBy( "email" )
    .clearOrders();

SELECT *
FROM `users`

reorder

query.from( "users" )
    .orderBy( "email" )
    .reorder( "username" );

SELECT *
FROM `users`
ORDER BY `username` ASC

Specifying A Select Clause

You may not always want to select all columns from a database table. You can influence the select list of a query with the following methods.

Individual columns can contain fully-qualified names (some_table.some_column), table aliases (alias.some_column), and even set column aliases themselves (some_column AS c). The columns argument can be a single column, a list of columns (comma-separated), or an array of columns.

select