Docs Menu

$sortArray (aggregation)

On this page

  • Definition
  • Behavior
  • Example
$sortArray

New in version 5.2.

Sorts an array based on its elements. The sort order is user specified.

$sortArray has the following syntax:

$sortArray: {
input: <array>,
sortBy: <sort spec>
}
Field
Type
Description
input
expression

The array to be sorted.

The result is null if the expression:

  • is missing
  • evaluates to null
  • evaluates to undefined

If the expression evaluates to any other non-array value, the document returns an error.

sortBy
document
The document specifies a sort ordering.

The $sortArray expression orders the input array according to the sortBy specification.

The $sortArray syntax and semantics are the same as the behavior in a $push operation modified by $sort.

If the array elements are documents, you can sort by a document field. Specify the field name and a sort direction, ascending (1), or descending (-1 ).

{
input: <array-of-documents>,
sortBy: { <document-field>: {sort-direction> }
}

To sort the whole array by value, or to sort by array elements that are not documents, identify the input array and specify 1 for an ascending sort or -1 for descending sort in the sortBy parameter.

{
input: <array-of-documents>,
sortBy: { sort-direction> }
}
  • There is no implicit array traversal on the sort key.
  • Positional operators are not supported. A field name like "values.1" denotes a sub-field called "1" in the values array. It does not refer to the item at index 1 in the values array.
  • When a whole array is sorted, the sort is lexicographic. The aggregation $sort stage, behaves differently. See $sort for more details.
  • When an array is sorted by a field, any documents or scalars that do not have the specified field are sorted equally. The resulting sort order is undefined.
  • null values and missing values sort equally.

The stability of the sort is not specified. Users should not rely on $sortArray to use a particular sorting algorithm.

Create the team collection:

db.engineers.insertOne(
{
"team":
[
{
"name": "pat",
"age": 30,
"address": { "street": "12 Baker St", "city": "London" }
},
{
"name": "dallas",
"age": 36,
"address": { "street": "12 Cowper St", "city": "Palo Alto" }
},
{
"name": "charlie",
"age": 42,
"address": { "street": "12 French St", "city": "New Brunswick" }
}
]
}
)

The team array has three elements. Each element of team has nested sub-elements: name, age, and address. The following examples show how to sort the team array using these sub-elements.

Sort on a field within an array element:

db.engineers.aggregate( [
{ $project:
{
_id: 0,
result:
{
$sortArray: { input: "$team", sortBy: { name: 1 } }
}
}
}
] )

The name field is a sub-element in the team array. The operation returns the following results:

{
result:
[
{
name: 'charlie',
age: 42,
address: { street: '12 French St', city: 'New Brunswick' }
},
{
name: 'dallas',
age: 36,
address: { street: '12 Cowper St', city: 'Palo Alto' }
},
{
name: 'pat',
age: 30,
address: { street: '12 Baker St', city: 'London' }
}
]
}

The address field is a document with subfields of its own. Use dot notation to sort the array based on a subfield:

db.engineers.aggregate( [
{
$project:
{
_id: 0,
result:
{
$sortArray:
{
input: "$team",
sortBy: { "address.city": -1 }
}
}
}
}
] )

The sort direction is descending because the sortBy value is "-1".

{
result: [
{
name: 'dallas',
age: 36,
address: { street: '12 Cowper St', city: 'Palo Alto' }
},
{
name: 'charlie',
age: 42,
address: { street: '12 French St', city: 'New Brunswick' }
},
{
name: 'pat',
age: 30,
address: { street: '12 Baker St', city: 'London' }
}
]
}

Specify multiple index fields to do a compound sort:

db.engineers.aggregate( [
{
$project:
{
_id: 0,
result:
{
$sortArray:
{
input: "$team",
sortBy: { age: -1, name: 1 }
}
}
}
}
] )

Example output:

{
name: 'charlie',
age: 42,
address: { street: '12 French St', city: 'New Brunswick' }
},
{
name: 'dallas',
age: 36,
address: { street: '12 Cowper St', city: 'Palo Alto' }
},
{
name: 'pat',
age: 30,
address: { street: '12 Baker St', city: 'London' }
}

This example specifies an input array directly. The values are all the same type, Int32:

db.engineers.aggregate( [
{
$project:
{
_id: 0,
result:
{
$sortArray:
{
input: [ 1, 4, 1, 6, 12, 5 ],
sortBy: 1
}
}
}
}
] )

Example output:

[ { result: [ 1, 1, 4, 5, 6, 12 ] } ]

The input array has a "1" in position 0 and position 2. The ones are grouped together in the results, but their are no guarantees regarding how the group of ones is sorted relative to their original order.

This example specifies an input array directly. The values have different types:

db.engineers.aggregate( [
{
$project:
{
_id: 0,
result:
{
$sortArray:
{
input: [ 20, 4, { a: "Free" }, 6, 21, 5, "Gratis",
{ a: null }, { a: { sale: true, price: 19 } },
Decimal128( "10.23" ), { a: "On sale" } ],
sortBy: 1
}
}
}
}
] )

Example output:

{ result: [
4,
5,
6,
Decimal128("10.23"),
20,
21,
'Gratis',
{ a: null },
{ a: 'Free' },
{ a: 'On sale' },
{ a: { sale: true, price: 19 } }
] }

The results are ordered.

In contrast, after changing the sortBy field to sort on the one of the document fields, sortBy: { a: 1 }, the sort order for the scalar and null values is undefined:

{ result: [
20,
4,
6,
21,
5,
'Gratis',
{ a: null },
Decimal128("10.23"),
{ a: 'Free' },
{ a: 'On sale' },
{ a: { sale: true, price: 19 } }
] }
←  $slice (aggregation)$split (aggregation) →
Give Feedback
© 2022 MongoDB, Inc.

About

  • Careers
  • Investor Relations
  • Legal Notices
  • Privacy Notices
  • Security Information
  • Trust Center
© 2022 MongoDB, Inc.