Docs Menu

Docs HomeDevelop ApplicationsAtlas Device SDKs

Read Realm Objects - Kotlin SDK

On this page

  • Read Operations
  • Frozen and Live Results
  • Find Database Objects
  • Find Latest Version of an Object
  • Query All Objects of a Type
  • Query a Single Object
  • Filter By Property
  • Filter By Primary Key
  • Filter By Embedded Object Property
  • Filter By RealmAny (Mixed) Property
  • Filter By Remapped Property
  • Filter By Full-Text Search (FTS) Property
  • Filter By Collection (List, Set, Dictionary) Property
  • Filter By Relationship Property
  • Query Geospatial Data
  • Sort and Limit Results
  • Aggregate Results
  • Iterate Results using Flow

This page describes how to query and lazily read objects persisted in a database with the Atlas Device SDK for Kotlin. This lazy evaluation enables code efficiency and performance when handling large data sets and complex queries.

A read operation consists of querying database objects, then running the query when you're ready to access the results. The syntax for read operations is the same for synced and non-synced databases.

All queries are based on object type. You can query any object, including embedded objects, that persist to the database and whose type is included in your database schema.

Construct queries using the SDK's query builder RealmQuery, passing the object type as the type parameter. You can query objects on a Realm or MutableRealm instance, a RealmResults collection, or a RealmList collection.

A basic RealmQuery returns all objects of the specified type:

// Finds all objects of type <T>
.query<T>()

You can refine your query with additional filters and conditions (for example, sorting, aggregating, limiting results). You can refine results by one or more property using Realm Query Language (RQL), a string-based query language, in combination with built-in Kotlin extension functions and helper methods provided by the SDK.

// Finds all objects of type <T> that meet the filter conditions
// and returns them in the specified sort order
.query<T>(filter).sort(sort)

You can also chain queries together using additional query() methods. Each appended query() acts as an AND query condition. And because of the SDK's lazy evaluation, successive queries do not require separate trips to the database.

// Finds all objects of type <T> that meet the first filter conditions
// AND the subsequent filter conditions
.query<T>(filter).query(filter)

When you're ready to access the data and work with the returned results, run the query:

  • Use find() to perform a synchronous query. The SDK lazily returns a RealmResults collection, which represents all database objects that match the query conditions. In general, you can can work with a results collection like any other Kotlin Collection.

  • Use asFlow() to perform an asynchronous query. The SDK lazily subscribes to a Kotlin Coroutine Flow that you can collect and iterate over or listen to for changes. You cannot call asFlow() on a MutableRealm.query.

Note that any retrieved results don't actually hold matching database objects in memory. Instead, the database uses direct references, or pointers. Database objects in a results collection or flow reference the matched objects, which map directly to data in the database file. This also means that you can traverse your graph of an object's relationships directly through the results of a query.

Example

Run the Query

val queryAllFrogs = realm.query<Frog>()
val queryAllLiveFrogs = this.query<Frog>() // this: MutableRealm
// Calling 'find()' on the query returns a RealmResults collection
// Can be called on a `Realm.query()` or `MutableRealm.query()`
val allFrogs: RealmResults<Frog> = queryAllFrogs.find()
val allLiveFrogs: RealmResults<Frog> = queryAllLiveFrogs.find()
// Calling 'asFlow()' on the query returns a ResultsChange Flow
// Can ONLY be called on a `Realm.query()`
val allFrogsFlow: Flow<ResultsChange<Frog>> = queryAllFrogs.asFlow()

Unlike other Atlas Device SDKs, which always return live results, results with the Kotlin SDK can be frozen or live. For more information on the Kotlin SDK's frozen architecture, refer to Frozen Architecture - Kotlin SDK.

To access frozen results, run a query on a Realm. Frozen results cannot be modified and do not reflect the latest changes to the database. A Realm.query() does not require a write transaction.

To access live results, run a query on a MutableRealm instance in a write transaction. A MutableRealm represents a writeable state of a database and is only accessible through a write transaction. Results from a MutableRealm.query are live, but are only valid on the calling thread and are frozen once the write transaction completes. For more information on write transactions and accessing a MutableRealm, refer to Write Transactions.

You can also access live objects from frozen results by calling MutableRealm.findLatest() . For more information, refer to the Find Latest Version of an Object section on this page.

Example

Access Live Results

// 'Realm.query()' results are always frozen
val frozenResults = realm.query<Frog>("age > $0", 50).find()
// 'MutableRealm.query()' results are live within the current write transaction
realm.write { // this: MutableRealm
val liveResults = this.query<Frog>("age > $0", 50).find()

To find objects stored within a database:

  1. Pass the object type as a type parameter to query(). The object type must already be included in your database schema.

  2. Optionally, pass any query conditions to further refine the results:

    • Specify a filter to only return objects that meet the condition. If you don't specify a filter, the SDK returns all objects of the specified type.

      You can chain filters by appending additional query() methods to the RealmQuery.

    • Specify the sort order for the results. Because the database is unordered, if you don't include a sort order, the SDK cannot guarantee the query returns objects in any specific order.

  3. Execute the query using either:

    • find() for synchronous queries. Returns a collection of results.

    • asFlow() f or asynchronous queries. Subscribes to a Flow of results changes.

    Tip

    Prefer asFlow() for Large Data Sets

    find() runs a synchronous query on the thread it is called from. As a result, avoid using find() for large data sets on the UI thread or in logic that could delay the UI thread.

    Prefer asFlow() to prevent negative performance or UI impact.

  4. Work with the results. Objects may be frozen or live, depending on the type of query you ran.

Example

Read Operation

// Pass the object type as <T> parameter and filter by property
val findFrogs = realm.query<Frog>("age > 1")
// Chain another query filter
.query("owner == $0 AND name CONTAINS $1", "Jim Henson", "K")
// Sort results by property
.sort("age", Sort.ASCENDING)
// Run the query
.find()
// ... work with the results

Because of the SDK's frozen architecture, you aren't always working with the latest version of an object or collection.

To get a version of an object or collection that reflects the latest changes to the database, you can call findLatest() from a MutableRealm instance. Like a MutableRealm.query(), the results are live but are only valid on the calling thread and are frozen once the write transaction completes.

In the following example, we pass an existing frozenFrogs query's RealmResults to findLatest() to get the latest live copy of the collection. We then modify the now-live objects within the write transaction:

// Open a write transaction to access the MutableRealm
realm.write { // this: MutableRealm
for (frog in frozenFrogs) {
// Call 'findLatest()' on an earlier query's frozen results
// to return the latest live objects that you can modify
// within the current write transaction
findLatest(frog)?.also { liveFrog ->
copyToRealm(liveFrog.apply { age += 1 })
println(liveFrog.name + " is now " + liveFrog.age + " years old")
}
}
}

Tip

You can check if an object is frozen with the isFrozen() method.

val isFrozen = frog.isFrozen()

To query all objects of a specific type, pass the RealmObject or EmbeddedRealmObject object type as a type parameter to query() without any query arguments. The SDK returns all objects of the specified type.

Note

Cannot Read Asymmetric Objects

You cannot read asymmetric objects because they are special write-only objects that do not persist to the database. For information on how to use asymmetric objects in your application, refer to Stream Data to Atlas - Kotlin SDK.

In the following example, we query all RealmObject objects of type Frog:

// Query all Frog objects in the database
val queryAllFrogs = realm.query<Frog>()
val allFrogs = queryAllFrogs.find()

In the following example, we query all EmbeddedRealmObject objects of type EmbeddedAddress:

val queryAllEmbeddedAddresses = realm.query<EmbeddedAddress>()
val allEmbeddedAddresses = queryAllEmbeddedAddresses.find()

You can also query an embedded object through its parent object. For more information, refer to the Filter By Embedded Object Property section on this page.

Tip

Once you find an embedded object, you can use the EmbeddedRealmObject.parent() method to access its parent:

val getParent = embeddedObject.parent<Contact>()

To find a single object of a specific object type, call first() on the query. When you run the query, the SDK returns the first object that matches the conditions or null.

In the following example, we query a Frog object type and return the first object:

val querySingleFrog = realm.query<Frog>().first()
val singleFrog = querySingleFrog.find()
if (singleFrog != null) {
println("${singleFrog.name} is a frog.")
} else {
println("No frogs found.")
}

You can filter a query by any property in the object type that persists in the database. This includes child properties, which you can refer to using dot notation.

To filter by property, you can pass Realm Query Language (RQL) filters and operators, use Kotlin's built-in extension methods or the SDK's convenience methods, or use a combination. For information on all currently supported RQL operators and syntax, refer to the Realm Query Language reference documentation.

In the following example, we query a Frog object type and filter by the name property:

val filterByProperty = realm.query<Frog>("name == $0", "Kermit")
val frogsNamedKermit = filterByProperty.find()

Primary keys are unique identifiers for objects in a database, which makes them useful for querying specific objects.

To filter by a specific primary key, pass the object type as a type parameter and query the primary key field for the desired value.

In the following example, we query a Frog object and filter by the primary key property _id:

val filterByPrimaryKey = realm.query<Frog>("_id == $0", PRIMARY_KEY_VALUE)
val findPrimaryKey = filterByPrimaryKey.find().first()

Tip

Device Sync Always Uses _id as Primary Key

If you use Atlas Device Sync, you can always query by the primary key field _id. This is because the Device Sync data model requires objects have a primary key named _id. For more information, refer to Model Data with Device Sync - Kotlin SDK.

Embedded objects act as nested data inside of a single specific parent object. You can query an embedded object directly or as a property on its parent object. For information on querying an embedded object directly, refer to the Query All Objects of a Type section on this page.

To find an embedded object through its parent object, pass the parent object type as a type parameter and filter by the embedded object property using dot notation.

In the following example, we have a Contact parent object that contains an embedded object property called address. We query the Contact object type against the embedded object's address.street property:

// Use dot notation to access the embedded object properties as if it
// were in a regular nested object
val filterEmbeddedObjectProperty =
realm.query<Contact>("address.street == '123 Pond St'")
// You can also access properties nested within the embedded object
val queryNestedProperty = realm.query<Contact>()
.query("address.propertyOwner.name == $0", "Mr. Frog")

A RealmAny (Mixed) property represents a polymorphic value that can hold any one of its supported data types at a particular moment. You can query a RealmAny property the same way you would any property.

In the following example, we query a RealmAny property called favoriteThing for a frog with a favorite thing of type Int:

val filterByRealmAnyInt = realm.query<Frog>("favoriteThing.@type == 'int'")
val findFrog = filterByRealmAnyInt.find().first()

Unlike other properties, you must extract a RealmAny property's stored value before you can work with it. To extract the value, use the SDK's getter method for the stored type. If you use the wrong getter for the type, the SDK throws an exception.

A best practice is to use a conditional expression to get the currently stored type with RealmAny.type(), then extract the value based on the type. For a full list of getter methods, refer to the RealmAny API reference.

In the following example, we extract the value using RealmAny.asInt() since we know the returned frog's favorite thing is an Int type value:

val frogsFavoriteThing = findFrog.favoriteThing // Int
// Using the correct getter method returns the value
val frogsFavoriteNumber = frogsFavoriteThing?.asInt()
println("${findFrog.name} likes the number $frogsFavoriteNumber")

Tip

Handle Polymorphism with Conditional Expressions

Use a conditional when expression to handle the possible inner value class of a given RealmAny property:

// Handle possible types with a 'when' statement
frogsFavoriteThings.forEach { realmAny ->
if (realmAny != null) {
when (realmAny.type) {
RealmAny.Type.INT -> {
val intValue = realmAny.asInt()
// Do something with intValue ...
}
RealmAny.Type.STRING -> {
val stringValue = realmAny.asString()
// Do something with stringValue ...
}
RealmAny.Type.OBJECT -> {
val objectValue = realmAny.asRealmObject(Frog::class)
// Do something with objectValue ...
}
// Handle other possible types...
else -> {
// Debug or perform a default action for unhandled types
Log.d("Unhandled type: ${realmAny.type}")
}
}
}
}

Once you have the currently stored value, you can work with it the same way you would another value of that type.

Note

Byte, Char, Int, Long, and Short values are converted internally to int64_t values. Keep this in mind when comparing, sorting, or aggregating RealmAny values of these types.

If your data model includes remapped property names, you can filter by both the Kotlin property name used in your code and the remapped property name that's persisted in the database.

In the following example, the Frog object has a property named species in the code that is remapped to latin_name in the database:

@PersistedName("latin_name")
var species: String? = null // Remapped property

In the database, we can filter by either property name and return the same results:

val filterByKotlinName = realm.query<Frog>("species == $0", "Muppetarium Amphibius")
val findSpecies = filterByKotlinName.find().first()
val filterByRemappedName = realm.query<Frog>("latin_name == $0", "Muppetarium Amphibius")
val find_latin_name = filterByRemappedName.find().first()
// Both queries return the same object
assertEquals(findSpecies, find_latin_name)

Changed in version 1.11.0: Support for prefix wildcard searches

If your data model includes a Full-Text Search (FTS) index property, you can filter by the property with the TEXT predicate. Words in the query are converted to tokens by a tokenizer using the following rules:

  • Tokens can only consist of characters from ASCII and the Latin-1 supplement (western languages). All other characters are considered whitespace.

  • Words split by a hyphen (-) are split into two tokens. For example, full-text splits into full and text.

  • Tokens are diacritics- and case-insensitive.

You can search for an entire word or phrase, or limit your results with the following characters:

  • Exclude results for a word by placing the - character in front of the word. For example, fiction -science would include all search results for fiction and exclude those that include the word science.

  • In Kotlin SDK version 1.11.0 and later, you can specify prefixes by placing the * character at the end of a word. For example, fict* would include all search results for fiction and fictitious. (The Kotlin SDK does not currently support suffix searches.)

The SDK returns a Boolean match for the specified query, instead of a relevance-based match.

In the following example, the Frog object type has an FTS index property called physicalDescription that we can filter by to find different types of frogs:

// Filter by FTS property value using 'TEXT'
// Find all frogs with "green" in the physical description
val onlyGreenFrogs =
realm.query<Frog>("physicalDescription TEXT $0", "green").find()
// Find all frogs with "green" but not "small" in the physical description
val onlyBigGreenFrogs =
realm.query<Frog>("physicalDescription TEXT $0", "green -small").find()
// Find all frogs with "muppet-" and "rain-" in the physical description
val muppetsInTheRain =
realm.query<Frog>("physicalDescription TEXT $0", "muppet* rain*").find()

Depending on how you define your object type, you might have properties that are defined as one of the following supported Collection Types:

  • RealmList

  • RealmSet

  • RealmDictionary

You can query these collection properties the same way you would any other property using RQL. You can also use Kotlin's built-in collection functions to filter, sort, and iterate over the results.

You can query and iterate through a RealmList property as you would a Kotlin List.

In the following example, we query a RealmList property called favoritePonds:

// Find frogs with a favorite pond
val allFrogs = query<Frog>().find()
val frogsWithFavoritePond = allFrogs.query("favoritePonds.@size > $0", 0).find()
// Check if the list contains a value
for (frog in frogsWithFavoritePond) {
val likesBigPond = frog.favoritePonds.any { pond -> pond.name == "Big Pond" }
if (likesBigPond) {
Log.v("${frog.name} likes Big Pond")
} else {
Log.v("${frog.name} does not like Big Pond")
}
}

You can query and iterate through a RealmSet property as you would a Kotlin Set.

In the following example, we query a RealmSet property called favoriteSnacks:

// Find frogs with flies and crickets as a favorite snack
val filterBySnackSet = query<RealmSet_Frog>("favoriteSnacks.name CONTAINS $0 AND favoriteSnacks.name CONTAINS $1", "Flies", "Crickets")
val potentialFrogs = filterBySnackSet.find()
// Check if the set contains a value
val frogsThatLikeWorms = potentialFrogs.filter { frog ->
val requiredSnacks = query<RealmSet_Snack>("name == $0", "Worms")
frog.favoriteSnacks.contains(requiredSnacks.find().first())
}
for (frog in frogsThatLikeWorms) {
Log.v("${frog.name} likes both Flies, Worms, and Crickets")
}

You can query and iterate through a RealmDictionary property as you would a Kotlin Map.

In the following example, we query a RealmDictionary property called favoritePondsByForest, which maps a String key (forest) to a String value (pond):

// Find frogs who have forests with favorite ponds
val frogs = realm.query<Frog>().find()
val frogsWithFavoritePonds = frogs.query("favoritePondsByForest.@count > $0", 1).find()
val thisFrog = frogsWithFavoritePonds.first()
// Iterate through the map and log each key-value pair
for (forestName in thisFrog.favoritePondsByForest.keys) {
val pondName = thisFrog.favoritePondsByForest[forestName]
Log.v("Forest: $forestName, Pond: $pondName")
}
// Check if the dictionary contains a key
if (thisFrog.favoritePondsByForest.containsKey("Hundred Acre Wood")) {
Log.v("${thisFrog.name}'s favorite pond in Hundred Acre Wood is ${thisFrog.favoritePondsByForest["Hundred Acre Wood"]}")
}
// Check if the dictionary contains a value
if (thisFrog.favoritePondsByForest.containsValue("Picnic Pond")) {
Log.v("${thisFrog.name} lists Picnic Pond as a favorite pond")
}

Depending on how you define your object type, you might have properties that reference another database object. This can be a to-one, to-many, or inverse relationship.

For information on filtering by relationship properties that reference an embedded object, refer to the Filter By Embedded Object Property section on this page.

A to-one relationship property maps to a single instance of another object type. You can filter by the relationship property using dot notation, the same way you would a nested object.

In the following example, the Frog object type has a property called favoritePond of type Pond:

// Find frogs who have a favorite pond
val allFrogs = query<Frog>().find()
val frogsWithFavoritePond = allFrogs.query("favoritePond.@count == $0", 1).find()
// Iterate through the results
for (frog in frogsWithFavoritePond) {
Log.v("${frog.name} likes ${frog.favoritePond?.name}")
}

To-many relationships properties are collections (a RealmList or RealmSet) of another object type. You can filter by and iterate through the relationship property the same way you would any other collection property.

In the following example, the Forest object type has a property called nearbyPonds that is a RealmList of type Pond:

// Find all forests with at least one nearby pond
val allForests = query<Forest>().find()
val forestsWithPonds = allForests.query("nearbyPonds.@count > $0", 0).find()
// Iterate through the results
for (forest in forestsWithPonds) {
val bigPond = query<Pond>("name == $0", "Big Pond").find().first()
if (forest.nearbyPonds.contains(bigPond)) {
Log.v("${forest.name} has a nearby pond named ${bigPond.name}")
} else {
Log.v("${forest.name} does not have a big pond nearby")
}
}

Unlike to-one and to-many relationships, an inverse relationship automatically creates a backlink between parent and child objects. This means that you can always query against both the parent and child. You can also use RQL-specific syntax to query the backlink (for more information, refer to Backlink Queries).

In the following example, a parent object of type User has an inverse relationship to a child object of type Post. We can query the parent object's User.posts relationship ("User has many Posts") as well as the inverse Post.user relationship ("Post belongs to User"):

// Query the parent object
val filterByUserName = query<User>("name == $0", "Kermit")
val kermit = filterByUserName.find().first()
// Use dot notation to access child objects
val myFirstPost = kermit.posts[0]
// Iterate through the backlink collection property
kermit.posts.forEach { post ->
Log.v("${kermit.name}'s Post: ${post.date} - ${post.title}")
}
// Filter posts through the parent's backlink property
// using `@links.<ObjectType>.<PropertyName>` syntax
val oldPostsByKermit = realm.query<Post>("date < $0", today)
.query("@links.User.posts.name == $0", "Kermit")
.find()
// Query the child object to access the parent
val post1 = query<Post>("title == $0", "Forest Life").find().first()
val post2 = query<Post>("title == $0", "Top Ponds of the Year!").find().first()
val parent = post1.user.first()

Important

Querying Inverse Relationship by Remapped Class Names

If the inverse relationship property is an object type with a remapped (persisted) class name, you must use the remapped class name in the raw RQL query.

@PersistedName(name = "Blog_Author")
class User : RealmObject {
@PrimaryKey
var _id: ObjectId = ObjectId()
var name: String = ""
var posts: RealmList<Post> = realmListOf()
}
// Filter by the remapped object type name
// using `@links.<RemappedObjectType>.<PropertyName>` syntax
val postsByKermit = realm.query<Post>()
.query("@links.Blog_Author.posts.name == $0", "Kermit")
.find()

New in version 1.11.0.

Kotlin SDK version 1.11.0 and later adds experimental geospatial APIs that support querying with geospatial data.

Note

To persist geospatial data, you must define a custom GeoJSON-compatible embedded class in your data model. For more information on the requirements, refer to Persist Geospatial Data.

Geospatial data is persisted as latitude/longitude pair in a custom embedded object's coordinates property. A geospatial query checks whether the point defined by the coordinates property is contained within the boundary of a defined geospatial shape.

The SDK supports the following geospatial shapes:

  • GeoCircle: defined by a center GeoPoint and a radius Distance

  • GeoBox: defined by two GeoPoint coordinates that represent the southwest and northeast corners of the box

  • GeoPolygon: defined by a set of at least 4 GeoPoint coordinates that represent a closed shape. This shape can contain holes that represent exclusive boundaries within the confines of the defined polygon.

For more information on geospatial shapes and how to define them, refer to Geospatial Data - Kotlin SDK.

To query geospatial data:

  1. Create an object containing the embedded geospatial data.

  2. Define the geospatial shape to set the boundary for the query.

  3. Query using the GEOWITHIN RQL operator. This method takes the coordinates property of your GeoJSON-compatible embedded object and checks if that point is contained within the boundary of a defined shape. The syntax is the same, regardless of the geospatial shape.

In the following example, we want to query for geospatial data persisted in a Company object through its embedded location property. We create two GeoCircle objects to set our query boundary:

val circle1 = GeoCircle.create(
center = GeoPoint.create(47.8, -122.6),
radius = Distance.fromKilometers(44.4)
)
val circle2 = GeoCircle.create(
center = GeoPoint.create(47.3, -121.9),
radius = Distance.fromDegrees(0.25)
)

We then query for any Company objects with a location contained within the defined GeoCircle boundaries:

val companiesInLargeCircle =
realm.query<Company>("location GEOWITHIN $circle1").find()
println("Companies in large circle: ${companiesInLargeCircle.size}")
val companiesInSmallCircle =
realm.query<Company>("location GEOWITHIN $circle2").find()
println("Companies in small circle: ${companiesInSmallCircle.size}")
Companies in large circle: 1
Companies in small circle: 0
Querying a GeoCircle example
click to enlarge

To ensure results are returned as expected, you can specify a sort order and limit conditions using RQL Sort, Limit, and Distinct operators, one of the following SDK convenience methods, or a combination of both:

Important

Order Matters

Regardless of whether you use RQL or convenience methods, the SDK executes each request in the order it's added to the query. This can impact the results returned. For example, sorting a query before limiting it can return very different results than sorting after limiting it.

In the following example, we sort and limit using convenience methods only, RQL only, and then a combination of both to return the same results:

// Query for all frogs owned by Jim Henson, then:
// 1. Sort results by age in descending order
// 2. Limit results to only distinct names
// 3. Limit results to only the first 2 objects
val organizedWithMethods = realm.query<Frog>("owner == $0", "Jim Henson")
.sort("age", Sort.DESCENDING)
.distinct("name")
.limit(2)
.find()
organizedWithMethods.forEach { frog ->
Log.v("Method sort: ${frog.name} is ${frog.age}")
}
val organizedWithRql = realm.query<Frog>()
.query("owner == $0 SORT(age DESC) DISTINCT(name) LIMIT(2)", "Jim Henson")
.find()
organizedWithRql.forEach { frog ->
Log.v("RQL sort: ${frog.name} is ${frog.age}")
}
val organizedWithBoth = realm.query<Frog>()
.query("owner == $0 SORT(age DESC)", "Jim Henson")
.distinct("name")
.limit(2)
.find()
organizedWithBoth.forEach { frog ->
Log.v("Combined sort: ${frog.name} is ${frog.age}")
}
Method sort: Kermit, Sr. is 100
Method sort: Kermit is 42
RQL sort: Kermit, Sr. is 100
RQL sort: Kermit is 42
Combined sort: Kermit, Sr. is 100
Combined sort: Kermit is 42

Note

String sorting and case-insensitive queries are only supported for character sets in 'Latin Basic', 'Latin Supplement', 'Latin Extended A', and 'Latin Extended B' (UTF-8 range 0-591).

You can also aggregate results, which reduces results to a single value based on a specified numerical property or collection. You can use RQL aggregate operators, one of the following convenience methods, or a combination of both:

In the following example, we aggregate the age property of a Frog object type:

val jimHensonsFrogs = realm.query<Frog>("owner == $0", "Jim Henson")
// Find the total number of frogs owned by Jim Henson
val numberOfJimsFrogs = jimHensonsFrogs.count().find()
// Find the oldest frog owned by Jim Henson
val maxAge = jimHensonsFrogs.max<Int>("age").find()
val oldestFrog = jimHensonsFrogs.query("age == $0", maxAge).find().first()

You can iterate through results using a Kotlin Coroutine Flow.

To convert query results into an asynchronous Flow, call asFlow() on the query. The SDK returns a ResultsChange Flow that you can iterate through with flow.collect().

In the following example, we iterate through a Flow of Frog objects:

// Get a Flow of all frogs in the database
val allFrogsQuery = realm.query<Frog>()
val frogsFlow: Flow<ResultsChange<Frog>> = allFrogsQuery.asFlow()
// Iterate through the Flow with 'collect()'
val frogsObserver: Deferred<Unit> = async {
frogsFlow.collect { results ->
when (results) {
is InitialResults<Frog> -> {
for (frog in results.list) {
Log.v("Frog: $frog")
}
}
else -> { /* no-op */ }
}
}
}
// ... Later, cancel the Flow, so you can safely close the database
frogsObserver.cancel()
realm.close()

Tip

Subscribe to Flows to Listen for Changes

After generating a Flow from a query, you can register a notification handler to listen for changes to the ResultsChanges. For more information, refer to React to Changes - Kotlin SDK.

← Create Realm Objects - Kotlin SDK