Supported Types - Swift SDK
On this page
- Collection Types
- Results and Sectioned Results
- Collections as Properties
- Collections are Live
- Supported Property Types
- Property Cheat Sheet
- Unique Identifiers
- Size Limitations
- AnyRealmCollection
- Mutable Set
- Map/Dictionary
- AnyRealmValue
- Collections as Mixed
- Geospatial Data
- Map Unsupported Types to Supported Types
- Declare Type Projections
- Conform to the Type Projection Protocol
- Use Type Projection in the Model
Collection Types
Realm has several types to represent groups of objects, which we call collections. A collection is an object that contains zero or more instances of one Realm type. Realm collections are homogenous: all objects in a collection are of the same type.
You can filter and sort any collection using Realm's query engine. Collections are live, so they always reflect the current state of the realm instance on the current thread. You can also listen for changes in the collection by subscribing to collection notifications.
All collection types conform to the RealmCollection protocol. This protocol inherits from CollectionType, so you can use a Realm collection as you would any other standard library collections.
Using the RealmCollection protocol, you can write generic code that can operate on any Realm collection:
func operateOn<C: RealmCollection>(collection: C) { // Collection could be either Results or List print("operating on collection containing \(collection.count) objects") }
Results and Sectioned Results
The Swift SDK Results collection is a class representing objects retrieved from queries. A Results collection represents the lazily-evaluated results of a query operation. Results are immutable: you cannot add or remove elements to or from the results collection. Results have an associated query that determines their contents.
The Swift SDK also provides SectionedResults,
a type-safe collection which holds ResultsSection
as its elements.
Each ResultSection is a results
collection that contains only objects that belong to a given section key.
For example, an app that includes a contact list might use SectionedResults
to display a list of contacts divided into sections, where each section
contains all the contacts whose first name starts with the given letter.
The ResultsSection
whose key is "L" would contain "Larry", "Liam",
and "Lisa".
Collections as Properties
The Swift SDK also offers several collection types you can use as properties in your data model:
List, a class representing to-many relationships in models.
LinkingObjects, a class representing inverse relationships in models.
MutableSet, a class representing a to-many relationship.
Map, a class representing an associative array of key-value pairs with unique keys.
AnyRealmCollection, a type-erased class that can forward calls to a concrete Realm collection like Results, List or LinkingObjects.
Collections are Live
Like live objects, Realm collections are usually live:
Live results collections always reflect the current results of the associated query.
Live lists always reflect the current state of the relationship on the realm instance.
There are two cases when a collection is not live:
The collection is unmanaged. For example, a List property of a Realm object that has not been added to a realm yet or that has been copied from a realm is not live.
The collection is frozen.
Combined with collection notifications, live collections enable clean, reactive code. For example, suppose your view displays the results of a query. You can keep a reference to the results collection in your view class, then read the results collection as needed without having to refresh it or validate that it is up-to-date.
Important
Results indexes may change
Since results update themselves automatically, do not store the positional index of an object in the collection or the count of objects in a collection. The stored index or count value could be outdated by the time you use it.
Supported Property Types
You can use the following types to define your object model properties.
To learn how specific data types are mapped to BSON types in an App Services Schema, refer to Data Model Mapping in the Atlas App Services documentation.
See also Model Data with Device Sync - Swift SDK.
Property Cheat Sheet
Changed in version 10.10.0: @Persisted
property declaration syntax
Type | Required | Optional | ||
---|---|---|---|---|
Bool |
|
| ||
Int, Int8, Int16, Int32, Int64 |
|
| ||
Float |
|
| ||
Double |
|
| ||
String |
|
| ||
Data |
|
| ||
Date |
|
| ||
Decimal128 |
|
| ||
|
| |||
|
| |||
| N/A | |||
| N/A | |||
| N/A | |||
| N/A | |||
User-defined Object | N/A |
| ||
User-defined EmbeddedObject | N/A |
| ||
User-defined Enums |
|
|
CGFloat
properties are discouraged, as the type is not
platform independent.
To use Key-Value Coding with a user-defined object in the @Persisted
syntax, add the @objc
attribute: @Persisted @objc var myObject: MyClass?
Setting Default Values
With the @Persisted
property declaration syntax, you may see a
performance impact when setting default values for:
List
MutableSet
Dictionary
Decimal128
UUID
ObjectId
@Persisted var listProperty: List<Int>
and @Persisted var
listProperty = List<Int>()
are both valid, and are functionally
equivalent. However, the second declaration will result in poorer
performance.
This is because the List is created when the parent object is created, rather than lazily as needed. For most types, this is a difference so small you can't measure it. For the types listed here, you may see a performance impact when using the second declaration style.
Type | Required | Optional | ||
---|---|---|---|---|
Boolean |
|
| ||
Integer |
|
| ||
Float |
|
| ||
Double |
|
| ||
String |
|
| ||
Data |
|
| ||
Date |
|
| ||
Decimal128 |
|
| ||
NSUUID |
|
| ||
|
| |||
| N/A | |||
| N/A | |||
| N/A | |||
User-defined RLMObject | N/A |
| ||
User-defined RLMEmbeddedObject | N/A |
|
Additionally:
Integral types
int
,NSInteger
,long
,long long
CGFloat
properties are discouraged, as the type is not
platform independent.
Changed in version 10.8.0: RealmProperty
replaces RealmOptional
Type | Required | Optional | ||
---|---|---|---|---|
Bool |
|
| ||
Int, Int8, Int16, Int32, Int64 |
|
| ||
Float |
|
| ||
Double |
|
| ||
String |
|
| ||
Data |
|
| ||
Date |
|
| ||
Decimal128 |
|
| ||
|
| |||
|
| |||
| ||||
| ||||
| ||||
| N/A | |||
User-defined Object | N/A |
|
Additionally:
EmbeddedObject-derived types
You can use RealmProperty <T?>
to
represent integers, doubles, and other types as optional.
CGFloat
properties are discouraged, as the type is not
platform independent.
Unique Identifiers
New in version 10.8.0: UUID
type
ObjectId
is a MongoDB-specific 12-byte unique value. UUID
is a
16-byte globally-unique value. You can index
both types, and use either as a primary key.
Note
When declaring default values for @Persisted
UUID or ObjectId property
attributes, both of these syntax types are valid:
@Persisted var value: UUID
@Persisted var value = UUID()
However, the second will result in poorer performance. This is because the latter creates a new identifier that is never used any time an object is read from the realm, while the former only creates them when needed.
@Persisted var id: ObjectId
has equivalent behavior to @objc dynamic
var _id = ObjectId.generate()
. They both make random ObjectIds.
@Persisted var _id = ObjectId()
has equivalent behavior to @objc
dynamic var _id = ObjectId()
. They both make zero-initialized ObjectIds.
Size Limitations
Data and string properties cannot hold more than 16MB. To store larger amounts of data, either:
Break the data into 16MB chunks, or
Store data directly on the file system and store paths to the files in the realm.
Realm throws a runtime exception if your app attempts to store more than 16MB in a single property.
To avoid size limitations and a performance impact, it is best not to store large blobs, such as image and video files, directly in a realm. Instead, save the file to a file store and keep only the location of the file and any relevant metadata in the realm.
AnyRealmCollection
To store a collection as a property or variable without needing to know the concrete collection type, Swift's type system requires a type-erased wrapper like AnyRealmCollection:
class ViewController { // let collection: RealmCollection // ^ // error: protocol 'RealmCollection' can only be used // as a generic constraint because it has Self or // associated type requirements // // init<C: RealmCollection>(collection: C) where C.ElementType == MyModel { // self.collection = collection // } let collection: AnyRealmCollection<MyModel> init<C: RealmCollection & _ObjcBridgeable>(collection: C) where C.ElementType == MyModel { self.collection = AnyRealmCollection(collection) } }
Mutable Set
New in version 10.8.0.
A MutableSet
collection represents a to-many relationship
containing distinct values. A MutableSet
supports the following types
(and their optional versions):
Bool
Data
Date
Decimal128
Double
Float
Int
Int8
Int16
Int32
Int64
Object
ObjectId
String
UUID
Like Swift's Set, MutableSet
is a
generic type that is parameterized on the type it stores. Unlike
native Swift collections,
Realm mutable sets are reference types, as opposed to value
types (structs).
You can only call the MutableSets
mutation methods during a write
transaction. As a result, MutableSets
are immutable if you open the
managing realm as a read-only realm.
You can filter and sort a MutableSet
with the same predicates as Results. Like other
Realm collections, you can register a change listener on a MutableSet
.
For example, a Dog
class model might contain a MutableSet
for
citiesVisited
:
class Dog: Object { var name = "" var currentCity = "" var citiesVisited: MutableSet<String> }
Note
When declaring default values for @Persisted
MutableSet property attributes,
both of these syntax types is valid:
@Persisted var value: MutableSet<String>
@Persisted var value = MutableSet<String>()
However, the second will result in significantly worse performance. This is because the MutableSet is created when the parent object is created, rather than lazily as needed.
Map/Dictionary
New in version 10.8.0.
The Map is an associative array that contains key-value pairs with unique keys.
Like Swift's Dictionary,
Map
is a generic type that is parameterized on its key and value
types. Unlike native Swift collections,
Realm Maps are reference types (classes), as opposed to
value types (structs).
You can declare a Map as a property of an object:
class Dog: Object { var name = "" var currentCity = "" // Map of city name -> favorite park in that city var favoriteParksByCity: Map<String, String> }
Realm disallows the use of .
or $
characters in map keys.
You can use percent encoding and decoding to store a map key that contains
one of these disallowed characters.
// Percent encode . or $ characters to use them in map keys let mapKey = "New York.Brooklyn" let encodedMapKey = "New York%2EBrooklyn"
Note
When declaring default values for @Persisted
Map property attributes, both
of these syntax types is valid:
@Persisted var value: Map<String, String>
@Persisted var value = Map<String, String>()
However, the second will result in significantly worse performance. This is because the Map is created when the parent object is created, rather than lazily as needed.
AnyRealmValue
Changed in version 10.51.0: AnyRealmValue
properties can hold lists or maps of mixed data.
New in version 10.8.0.
AnyRealmValue
is a Realm property type that can hold different
data types. Supported AnyRealmValue
data types include:
Int
Float
Double
Decimal128
ObjectID
UUID
Bool
Date
Data
String
List
Map
Object
AnyRealmValue
cannot hold a MutableSet
or embedded object.
This mixed data type
is indexable, but you can't use it as a
primary key. Because null
is a
permitted value, you can't declare an AnyRealmValue
as optional.
class Dog: Object { var name = "" var currentCity = "" var companion: AnyRealmValue }
Collections as Mixed
In version 10.51.0 and later, a AnyRealmValue
data type can
contain collections (a list or map, but not a set) of AnyRealmValue
elements. You can use mixed collections to
model unstructured or variable data. For more information, refer to
Define Unstructured Data.
You can nest mixed collections up to 100 levels.
You can query mixed collection properties and register a listener for changes, as you would a normal collection.
You can find and update individual mixed collection elements
You cannot store sets or embedded objects in mixed collections.
To use mixed collections in your app, define the AnyRealmValue
type
property in your data model.
Then, you can create the list or map collections like any other mixed data value.
Geospatial Data
New in version 10.47.0.
Geospatial data, or "geodata", specifies points and geometric objects on the Earth's surface.
If you want to persist geospatial data, it must conform to the GeoJSON spec.
To persist geospatial data with the Swift SDK, create a GeoJSON-compatible embedded class that you can use in your data model.
Your custom embedded object must contain the two fields required by the GeoJSON spec:
A field of type
String
property that maps to atype
property with the value of"Point"
:@Persisted var type: String = "Point"
A field of type
List<Double>
that maps to acoordinates
property containing a latitude/longitude pair:@Persisted private var coordinates: List<Double>
class CustomGeoPoint: EmbeddedObject { private var type: String = "Point" private var coordinates: List<Double> public var latitude: Double { return coordinates[1] } public var longitude: Double { return coordinates[0] } convenience init(_ latitude: Double, _ longitude: Double) { self.init() // Longitude comes first in the coordinates array of a GeoJson document coordinates.append(objectsIn: [longitude, latitude]) } }
Map Unsupported Types to Supported Types
New in version 10.20.0.
You can use Type Projection to persist unsupported types as supported types
in Realm. This enables you to work with Swift types that Realm
does not support, but store them as types that Realm does support. You could
store a URL as a String
, for example, but read it from
Realm and use it in your application as though it were a URL.
Declare Type Projections
To use type projection with Realm:
Use one of Realm's custom type protocols to map an unsupported data type to a type that Realm supports
Use the projected types as @Persisted properties in the Realm object model
Conform to the Type Projection Protocol
You can map an unsupported data type to a type that Realm supports using one of the Realm type projection protocols.
The Swift SDK provides two type projection protocols:
CustomPersistable
FailableCustomPersistable
Use CustomPersistable when there is no chance the conversion can fail.
Use FailableCustomPersistable when it is possible for the conversion to fail.
// Extend a type as a CustomPersistable if if is impossible for // conversion between the mapped type and the persisted type to fail. extension CLLocationCoordinate2D: CustomPersistable { // Define the storage object that is persisted to the database. // The `PersistedType` must be a type that Realm supports. // In this example, the PersistedType is an embedded object. public typealias PersistedType = Location // Construct an instance of the mapped type from the persisted type. // When reading from the database, this converts the persisted type to the mapped type. public init(persistedValue: PersistedType) { self.init(latitude: persistedValue.latitude, longitude: persistedValue.longitude) } // Construct an instance of the persisted type from the mapped type. // When writing to the database, this converts the mapped type to a persistable type. public var persistableValue: PersistedType { Location(value: [self.latitude, self.longitude]) } } // Extend a type as a FailableCustomPersistable if it is possible for // conversion between the mapped type and the persisted type to fail. // This returns nil on read if the underlying column contains nil or // something that can't be converted to the specified type. extension URL: FailableCustomPersistable { // Define the storage object that is persisted to the database. // The `PersistedType` must be a type that Realm supports. public typealias PersistedType = String // Construct an instance of the mapped type from the persisted type. // When reading from the database, this converts the persisted type to the mapped type. // This must be a failable initilizer when the conversion may fail. public init?(persistedValue: String) { self.init(string: persistedValue) } // Construct an instance of the persisted type from the mapped type. // When writing to the database, this converts the mapped type to a persistable type. public var persistableValue: String { self.absoluteString } }
Supported PersistedTypes
The PersistedType
can use any of the primitive types that the
Swift SDK supports. It can also be
an Embedded Object.
PersistedType
cannot be an optional or a collection. However you can use the mapped type as an
optional or collection property in your object model.
extension URL: FailableCustomPersistable { // The `PersistedType` cannot be an optional, so this is not a valid // conformance to the FailableCustomPersistable protocol. public typealias PersistedType = String? ... } class Club: Object { var id: ObjectId var name: String // Although the `PersistedType` cannot be optional, you can use the // custom-mapped type as an optional in your object model. var url: URL? }
Use Type Projection in the Model
A type that conforms to one of the type projection protocols can be used with
the @Persisted
property declaration syntax introduced in Swift SDK
version 10.10.0. It does not work with the @objc dynamic
syntax.
You can use projected types for:
Top-level types
Optional versions of the type
The types for a collection
When using a FailableCustomPersistable
as a property, define it as an
optional property. When it is optional, the FailableCustomPersistable
protocol maps invalid values to nil
. When it is a required property, it is
force-unwrapped. If you have a value that can't be converted to the projected
type, reading that property throws an unwrapped fail exception.
class Club: Object { var id: ObjectId var name: String // Since we declared the URL as a FailableCustomPersistable, // it must be optional. var url: URL? // Here, the `location` property maps to an embedded object. // We can declare the property as required. // If the underlying field contains nil, this becomes // a default-constructed instance of CLLocationCoordinate // with field values of `0`. var location: CLLocationCoordinate2D } public class Location: EmbeddedObject { var latitude: Double var longitude: Double }
When your model contains projected types, you can create the object with values using the persisted type, or by assigning to the field properties of an initialized object using the projected types.
// Initialize objects and assign values let club = Club(value: ["name": "American Kennel Club", "url": "https://akc.org"]) let club2 = Club() club2.name = "Continental Kennel Club" // When assigning the value to a type-projected property, type safety // checks for the mapped type - not the persisted type. club2.url = URL(string: "https://ckcusa.com/")! club2.location = CLLocationCoordinate2D(latitude: 40.7509, longitude: 73.9777)
Type Projection in the Schema
When you declare your type as conforming to a type projection protocol, you
specify the type that should be persisted in realm. For example, if
you map a custom type URL
to a persisted type of String
, a URL
property appears as a String
in the schema, and dynamic access to the
property acts on strings.
The schema does not directly represent mapped types. Changing a property from its persisted type to its mapped type, or vice versa, does not require a migration.