WorkManager
abstract class WorkManager
WorkManager is the recommended library for persistent work. Scheduled work is guaranteed to execute sometime after its Constraints are met. WorkManager allows observation of work status and the ability to create complex chains of work.
WorkManager uses an underlying job dispatching service when available based on the following criteria:
-
Uses JobScheduler for API 23+
-
Uses a custom AlarmManager + BroadcastReceiver implementation for API 14-22
All work must be done in a ListenableWorker class. A simple implementation, Worker, is recommended as the starting point for most developers. With the optional dependencies, you can also use CoroutineWorker or RxWorker. All background work is given a maximum of ten minutes to finish its execution. After this time has expired, the worker will be signalled to stop.
There are two types of work supported by WorkManager: OneTimeWorkRequest and PeriodicWorkRequest. You can enqueue requests using WorkManager as follows:
WorkManager workManager = WorkManager.getInstance(Context);
workManager.enqueue(new OneTimeWorkRequest.Builder(FooWorker.class).build());
A WorkRequest has an associated id that can be used for lookups and observation as follows:
WorkRequest request = new OneTimeWorkRequest.Builder(FooWorker.class).build();
workManager.enqueue(request);
LiveData<WorkInfo> status = workManager.getWorkInfoByIdLiveData(request.getId());
status.observe(...);
You can also use the id for cancellation:
WorkRequest request = new OneTimeWorkRequest.Builder(FooWorker.class).build();
workManager.enqueue(request);
workManager.cancelWorkById(request.getId());
You can chain work as follows:
WorkRequest request1 = new OneTimeWorkRequest.Builder(FooWorker.class).build();
WorkRequest request2 = new OneTimeWorkRequest.Builder(BarWorker.class).build();
WorkRequest request3 = new OneTimeWorkRequest.Builder(BazWorker.class).build();
workManager.beginWith(request1, request2).then(request3).enqueue();
Each call to beginWith returns a WorkContinuation upon which you can call WorkContinuation.then with a single OneTimeWorkRequest or a list of OneTimeWorkRequest to chain further work. This allows for creation of complex chains of work. For example, to create a chain like this:
A
|
+----------+
| |
B C
|
+----+
| |
D E
you would enqueue them as follows:
WorkContinuation continuation = workManager.beginWith(A);
continuation.then(B).then(D, E).enqueue(); // A is implicitly enqueued here
continuation.then(C).enqueue();
Work is eligible for execution when all of its prerequisites are complete. If any of its prerequisites fail or are cancelled, the work will never run.
WorkRequests can accept Constraints, inputs (see Data), and backoff criteria. WorkRequests can be tagged with human-readable Strings (see WorkRequest.Builder.addTag), and chains of work can be given a uniquely-identifiable name (see beginUniqueWork).
Initializing WorkManager
By default, WorkManager auto-initializes itself using a built-in ContentProvider. ContentProviders are created and run before the Application object, so this allows the WorkManager singleton to be setup before your code can run in most cases. This is suitable for most developers. However, you can provide a custom Configuration by using Configuration.Provider or WorkManager.initialize.
Renaming and Removing ListenableWorker Classes
Exercise caution in renaming classes derived from ListenableWorkers. WorkManager stores the class name in its internal database when the WorkRequest is enqueued so it can later create an instance of that worker when constraints are met. Unless otherwise specified in the WorkManager Configuration, this is done in the default WorkerFactory which tries to reflectively create the ListenableWorker object. Therefore, renaming or removing these classes is dangerous - if there is pending work with the given class, it will fail permanently if the class cannot be found. If you are using a custom WorkerFactory, make sure you properly handle cases where the class is not found so that your code does not crash.
In case it is desirable to rename a class, implement a custom WorkerFactory that instantiates the right ListenableWorker for the old class name.
Summary
Nested types |
|---|
enum WorkManager.UpdateResult : EnumAn enumeration of results for |
Public companion functions |
|
|---|---|
open WorkManager |
This function is deprecated. Use the overload receiving Context |
open WorkManager |
getInstance(context: Context)Retrieves the |
open Unit |
initialize(context: Context, configuration: Configuration)Used to do a one-time initialization of the |
open Boolean |
Provides a way to check if |
Public functions |
|
|---|---|
WorkContinuation |
beginUniqueWork(This method allows you to begin unique chains of work for situations where you only want one chain with a given name to be active at a time. |
abstract WorkContinuation |
beginUniqueWork(This method allows you to begin unique chains of work for situations where you only want one chain with a given name to be active at a time. |
WorkContinuation |
beginWith(request: OneTimeWorkRequest)Begins a chain with one or more |
abstract WorkContinuation |
beginWith(requests: List<OneTimeWorkRequest>)Begins a chain with one or more |
abstract Operation |
Cancels all unfinished work. |
abstract Operation |
cancelAllWorkByTag(tag: String)Cancels all unfinished work with the given tag. |
abstract Operation |
cancelUniqueWork(uniqueWorkName: String)Cancels all unfinished work in the work chain with the given name. |
abstract Operation |
cancelWorkById(id: UUID)Cancels work with the given id if it isn't finished. |
abstract PendingIntent |
Creates a |
Operation |
enqueue(request: WorkRequest)Enqueues one item for background processing. |
abstract Operation |
enqueue(requests: List<WorkRequest>)Enqueues one or more items for background processing. |
abstract Operation |
enqueueUniquePeriodicWork(This method allows you to enqueue a uniquely-named |
open Operation |
enqueueUniqueWork(This method allows you to enqueue |
abstract Operation |
enqueueUniqueWork(This method allows you to enqueue |
abstract ListenableFuture<Long> |
Gets a |
abstract LiveData<Long> |
Gets a |
abstract ListenableFuture<WorkInfo?> |
getWorkInfoById(id: UUID)Gets a |
abstract Flow<WorkInfo?> |
getWorkInfoByIdFlow(id: UUID) |
abstract LiveData<WorkInfo?> |
|
abstract ListenableFuture<List<WorkInfo>> |
getWorkInfos(workQuery: WorkQuery)Gets the |
abstract ListenableFuture<List<WorkInfo>> |
getWorkInfosByTag(tag: String)Gets a |
abstract Flow<List<WorkInfo>> |
getWorkInfosByTagFlow(tag: String) |
abstract LiveData<List<WorkInfo>> |
Gets a |
abstract Flow<List<WorkInfo>> |
getWorkInfosFlow(workQuery: WorkQuery)Gets the |
abstract ListenableFuture<List<WorkInfo>> |
getWorkInfosForUniqueWork(uniqueWorkName: String)Gets a |
abstract Flow<List<WorkInfo>> |
getWorkInfosForUniqueWorkFlow(uniqueWorkName: String)Gets a |
abstract LiveData<List<WorkInfo>> |
getWorkInfosForUniqueWorkLiveData(uniqueWorkName: String)Gets a |
abstract LiveData<List<WorkInfo>> |
getWorkInfosLiveData(workQuery: WorkQuery)Gets the |
abstract Operation |
Prunes all eligible finished work from the internal database. |
abstract ListenableFuture<WorkManager.UpdateResult> |
updateWork(request: WorkRequest)Updates the work with the new specification. |
Public properties |
|
|---|---|
abstract Configuration |
The |
Public companion functions
open fungetInstance(): WorkManager
Retrieves the default singleton instance of WorkManager.
| Returns | |
|---|---|
WorkManager |
The singleton instance of |
| Throws | |
|---|---|
kotlin.IllegalStateException |
If WorkManager is not initialized properly as per the exception message. |
getInstance
open fun getInstance(context: Context): WorkManager
Retrieves the default singleton instance of WorkManager.
| Returns | |
|---|---|
WorkManager |
The singleton instance of |
| Throws | |
|---|---|
kotlin.IllegalStateException |
If WorkManager is not initialized properly |
initialize
open fun initialize(context: Context, configuration: Configuration): Unit
Used to do a one-time initialization of the WorkManager singleton with a custom Configuration. By default, this method should not be called because WorkManager is automatically initialized. To initialize WorkManager yourself, please follow these steps:
-
Disable
androidx.work.WorkManagerInitializerin your manifest. -
Invoke this method in
Application#onCreateor aContentProvider. Note that this method must be invoked in one of these two places or you risk getting aNullPointerExceptioningetInstance.
This method throws an IllegalStateException when attempting to initialize in direct boot mode.
This method throws an exception if it is called multiple times.
| Parameters | |
|---|---|
context: Context |
A |
configuration: Configuration |
The |
| See also | |
|---|---|
Configuration.Provider |
for on-demand initialization. |
isInitialized
open fun isInitialized(): Boolean
Provides a way to check if WorkManager is initialized in this process.
| Returns | |
|---|---|
Boolean |
|
Public functions
beginUniqueWork
fun beginUniqueWork(
uniqueWorkName: String,
existingWorkPolicy: ExistingWorkPolicy,
request: OneTimeWorkRequest
): WorkContinuation
This method allows you to begin unique chains of work for situations where you only want one chain with a given name to be active at a time. For example, you may only want one sync operation to be active. If there is one pending, you can choose to let it run or replace it with your new work.
The uniqueWorkName uniquely identifies this set of work.
If this method determines that new work should be enqueued and run, all records of previous work with uniqueWorkName will be pruned. If this method determines that new work should NOT be run, then the entire chain will be considered a no-op.
If any work in the chain fails or is cancelled, all of its dependent work inherits that state and will never run. This is particularly important if you are using APPEND as your ExistingWorkPolicy.
| Parameters | |
|---|---|
uniqueWorkName: String |
A unique name which for this chain of work |
existingWorkPolicy: ExistingWorkPolicy |
|
request: OneTimeWorkRequest |
The |
| Returns | |
|---|---|
WorkContinuation |
A |
beginUniqueWork
abstract fun beginUniqueWork(
uniqueWorkName: String,
existingWorkPolicy: ExistingWorkPolicy,
requests: List<OneTimeWorkRequest>
): WorkContinuation
This method allows you to begin unique chains of work for situations where you only want one chain with a given name to be active at a time. For example, you may only want one sync operation to be active. If there is one pending, you can choose to let it run or replace it with your new work.
The uniqueWorkName uniquely identifies this set of work.
If this method determines that new work should be enqueued and run, all records of previous work with uniqueWorkName will be pruned. If this method determines that new work should NOT be run, then the entire chain will be considered a no-op.
If any work in the chain fails or is cancelled, all of its dependent work inherits that state and will never run. This is particularly important if you are using APPEND as your ExistingWorkPolicy.
| Parameters | |
|---|---|
uniqueWorkName: String |
A unique name which for this chain of work |
existingWorkPolicy: ExistingWorkPolicy |
An |
requests: List<OneTimeWorkRequest> |
One or more |
| Returns | |
|---|---|
WorkContinuation |
A |
beginWith
fun beginWith(request: OneTimeWorkRequest): WorkContinuation
Begins a chain with one or more OneTimeWorkRequests, which can be enqueued together in the future using WorkContinuation.enqueue.
If any work in the chain fails or is cancelled, all of its dependent work inherits that state and will never run.
| Parameters | |
|---|---|
request: OneTimeWorkRequest |
One or more |
| Returns | |
|---|---|
WorkContinuation |
A |
beginWith
abstract fun beginWith(requests: List<OneTimeWorkRequest>): WorkContinuation
Begins a chain with one or more OneTimeWorkRequests, which can be enqueued together in the future using WorkContinuation.enqueue.
If any work in the chain fails or is cancelled, all of its dependent work inherits that state and will never run.
| Parameters | |
|---|---|
requests: List<OneTimeWorkRequest> |
One or more |
| Returns | |
|---|---|
WorkContinuation |
A |
cancelAllWork
abstract fun cancelAllWork(): Operation
Cancels all unfinished work. Use this method with extreme caution! By invoking it, you will potentially affect other modules or libraries in your codebase. It is strongly recommended that you use one of the other cancellation methods at your disposal.
Upon cancellation, ListenableFuture returned by ListenableWorker.startWork will be cancelled. Also ListenableWorker.onStopped will be invoked for any affected workers.
cancelAllWorkByTag
abstract fun cancelAllWorkByTag(tag: String): Operation
Cancels all unfinished work with the given tag. Note that cancellation is a best-effort policy and work that is already executing may continue to run. Upon cancellation, ListenableFuture returned by ListenableWorker.startWork will be cancelled. Also ListenableWorker.onStopped will be invoked for any affected workers.
| Parameters | |
|---|---|
tag: String |
The tag used to identify the work |
cancelUniqueWork
abstract fun cancelUniqueWork(uniqueWorkName: String): Operation
Cancels all unfinished work in the work chain with the given name. Note that cancellation is a best-effort policy and work that is already executing may continue to run. Upon cancellation, ListenableFuture returned by ListenableWorker.startWork will be cancelled. Also ListenableWorker.onStopped will be invoked for any affected workers.
| Parameters | |
|---|---|
uniqueWorkName: String |
The unique name used to identify the chain of work |
cancelWorkById
abstract fun cancelWorkById(id: UUID): Operation
Cancels work with the given id if it isn't finished. Note that cancellation is a best-effort policy and work that is already executing may continue to run. Upon cancellation, ListenableFuture returned by ListenableWorker.startWork will be cancelled. Also ListenableWorker.onStopped will be invoked for any affected workers.
| Parameters | |
|---|---|
id: UUID |
The id of the work |
createCancelPendingIntent
abstract fun createCancelPendingIntent(id: UUID): PendingIntent
Creates a PendingIntent which can be used to cancel a WorkRequest with the given id.
| Parameters | |
|---|---|
id: UUID |
The |
| Returns | |
|---|---|
PendingIntent |
The |
enqueue
fun enqueue(request: WorkRequest): Operation
Enqueues one item for background processing.
| Parameters | |
|---|---|
request: WorkRequest |
The |
enqueue
abstract fun enqueue(requests: List<WorkRequest>): Operation
Enqueues one or more items for background processing.
| Parameters | |
|---|---|
requests: List<WorkRequest> |
One or more |
enqueueUniquePeriodicWork
abstract fun enqueueUniquePeriodicWork(
uniqueWorkName: String,
existingPeriodicWorkPolicy: ExistingPeriodicWorkPolicy,
request: PeriodicWorkRequest
): Operation
This method allows you to enqueue a uniquely-named PeriodicWorkRequest, where only one PeriodicWorkRequest of a particular name can be active at a time. For example, you may only want one sync operation to be active. If there is one pending, you can choose to let it run or replace it with your new work.
The uniqueWorkName uniquely identifies this PeriodicWorkRequest.
| Parameters | |
|---|---|
uniqueWorkName: String |
A unique name which for this operation |
existingPeriodicWorkPolicy: ExistingPeriodicWorkPolicy |
|
request: PeriodicWorkRequest |
A |
enqueueUniqueWork
open fun enqueueUniqueWork(
uniqueWorkName: String,
existingWorkPolicy: ExistingWorkPolicy,
request: OneTimeWorkRequest
): Operation
This method allows you to enqueue work requests to a uniquely-named WorkContinuation, where only one continuation of a particular name can be active at a time. For example, you may only want one sync operation to be active. If there is one pending, you can choose to let it run or replace it with your new work.
The uniqueWorkName uniquely identifies this WorkContinuation.
| Parameters | |
|---|---|
uniqueWorkName: String |
A unique name which for this operation |
existingWorkPolicy: ExistingWorkPolicy |
An |
request: OneTimeWorkRequest |
The |
enqueueUniqueWork
abstract fun enqueueUniqueWork(
uniqueWorkName: String,
existingWorkPolicy: ExistingWorkPolicy,
requests: List<OneTimeWorkRequest>
): Operation
This method allows you to enqueue work requests to a uniquely-named WorkContinuation, where only one continuation of a particular name can be active at a time. For example, you may only want one sync operation to be active. If there is one pending, you can choose to let it run or replace it with your new work.
The uniqueWorkName uniquely identifies this WorkContinuation.
| Parameters | |
|---|---|
uniqueWorkName: String |
A unique name which for this operation |
existingWorkPolicy: ExistingWorkPolicy |
|
requests: List<OneTimeWorkRequest> |
|
getLastCancelAllTimeMillis
abstract fun getLastCancelAllTimeMillis(): ListenableFuture<Long>
Gets a ListenableFuture of the last time all work was cancelled. This method is intended for use by library and module developers who have dependent data in their own repository that must be updated or deleted in case someone cancels their work without their prior knowledge.
| Returns | |
|---|---|
ListenableFuture<Long> |
A |
getLastCancelAllTimeMillisLiveData
abstract fun getLastCancelAllTimeMillisLiveData(): LiveData<Long>
Gets a LiveData of the last time all work was cancelled. This method is intended for use by library and module developers who have dependent data in their own repository that must be updated or deleted in case someone cancels their work without their prior knowledge.
| Returns | |
|---|---|
LiveData<Long> |
A |
getWorkInfoById
abstract fun getWorkInfoById(id: UUID): ListenableFuture<WorkInfo?>
Gets a ListenableFuture of the WorkInfo for a given work id.
| Parameters | |
|---|---|
id: UUID |
The id of the work |
| Returns | |
|---|---|
ListenableFuture<WorkInfo?> |
A |
getWorkInfoByIdFlow
abstract fun getWorkInfoByIdFlow(id: UUID): Flow<WorkInfo?>
Gets a Flow of the WorkInfo for a given work id.
| Parameters | |
|---|---|
id: UUID |
The id of the work |
getWorkInfoByIdLiveData
abstract fun getWorkInfoByIdLiveData(id: UUID): LiveData<WorkInfo?>
Gets a LiveData of the WorkInfo for a given work id.
| Parameters | |
|---|---|
id: UUID |
The id of the work |
getWorkInfos
abstract fun getWorkInfos(workQuery: WorkQuery): ListenableFuture<List<WorkInfo>>
Gets the ListenableFuture of the List of WorkInfo for all work referenced by the WorkQuery specification.
| Parameters | |
|---|---|
workQuery: WorkQuery |
The work query specification |
| Returns | |
|---|---|
ListenableFuture<List<WorkInfo>> |
A |
getWorkInfosByTag
abstract fun getWorkInfosByTag(tag: String): ListenableFuture<List<WorkInfo>>
Gets a ListenableFuture of the WorkInfo for all work for a given tag.
| Parameters | |
|---|---|
tag: String |
The tag of the work |
| Returns | |
|---|---|
ListenableFuture<List<WorkInfo>> |
A |
getWorkInfosByTagFlow
abstract fun getWorkInfosByTagFlow(tag: String): Flow<List<WorkInfo>>
Gets a Flow of the WorkInfo for all work for a given tag.
| Parameters | |
|---|---|
tag: String |
The tag of the work |
getWorkInfosByTagLiveData
abstract fun getWorkInfosByTagLiveData(tag: String): LiveData<List<WorkInfo>>
Gets a LiveData of the WorkInfo for all work for a given tag.
| Parameters | |
|---|---|
tag: String |
The tag of the work |
getWorkInfosFlow
abstract fun getWorkInfosFlow(workQuery: WorkQuery): Flow<List<WorkInfo>>
Gets the Flow of the List of WorkInfo for all work referenced by the WorkQuery specification.
| Parameters | |
|---|---|
workQuery: WorkQuery |
The work query specification |
getWorkInfosForUniqueWork
abstract fun getWorkInfosForUniqueWork(uniqueWorkName: String): ListenableFuture<List<WorkInfo>>
Gets a ListenableFuture of the WorkInfo for all work in a work chain with a given unique name.
| Parameters | |
|---|---|
uniqueWorkName: String |
The unique name used to identify the chain of work |
| Returns | |
|---|---|
ListenableFuture<List<WorkInfo>> |
A |
getWorkInfosForUniqueWorkFlow
abstract fun getWorkInfosForUniqueWorkFlow(uniqueWorkName: String): Flow<List<WorkInfo>>
Gets a Flow of the WorkInfo for all work in a work chain with a given unique name.
| Parameters | |
|---|---|
uniqueWorkName: String |
The unique name used to identify the chain of work |
getWorkInfosForUniqueWorkLiveData
abstract fun getWorkInfosForUniqueWorkLiveData(uniqueWorkName: String): LiveData<List<WorkInfo>>
Gets a LiveData of the WorkInfo for all work in a work chain with a given unique name.
| Parameters | |
|---|---|
uniqueWorkName: String |
The unique name used to identify the chain of work |
getWorkInfosLiveData
abstract fun getWorkInfosLiveData(workQuery: WorkQuery): LiveData<List<WorkInfo>>
Gets the LiveData of the List of WorkInfo for all work referenced by the WorkQuery specification.
| Parameters | |
|---|---|
workQuery: WorkQuery |
The work query specification |
pruneWork
abstract fun pruneWork(): Operation
Prunes all eligible finished work from the internal database. Eligible work must be finished (WorkInfo.State.SUCCEEDED, WorkInfo.State.FAILED, or WorkInfo.State.CANCELLED), with zero unfinished dependents.
Use this method with caution; by invoking it, you (and any modules and libraries in your codebase) will no longer be able to observe the WorkInfo of the pruned work. You do not normally need to call this method - WorkManager takes care to auto-prune its work after a sane period of time. This method also ignores the OneTimeWorkRequest.Builder.keepResultsForAtLeast policy.
updateWork
abstract fun updateWork(request: WorkRequest): ListenableFuture<WorkManager.UpdateResult>
Updates the work with the new specification. A WorkRequest passed as parameter must have an id set with WorkRequest.Builder.setId that matches an id of the previously enqueued work.
It preserves enqueue time, e.g. if a work was enqueued 3 hours ago and had 6 hours long initial delay, after the update it would be still eligible for run in 3 hours, assuming that initial delay wasn't updated.
If the work being updated is currently running the returned ListenableFuture will be completed with UpdateResult.APPLIED_FOR_NEXT_RUN. In this case the current run won't be interrupted and will continue to rely on previous state of the request, e.g. using old constraints, tags etc. However, on the next run, e.g. retry of one-time Worker or another iteration of periodic worker, the new worker specification will be used.
If the one time work that is updated is already finished the returned ListenableFuture will be completed with UpdateResult.NOT_APPLIED.
If update can be applied immediately, e.g. the updated work isn't currently running, the returned ListenableFuture will be completed with UpdateResult.APPLIED_IMMEDIATELY.
If the work with the given id (request.getId()) doesn't exist the returned ListenableFuture will be completed exceptionally with IllegalArgumentException.
Worker type can't be changed, OneTimeWorkRequest can't be updated to PeriodicWorkRequest and otherwise, the returned ListenableFuture will be completed with IllegalArgumentException.
| Parameters | |
|---|---|
request: WorkRequest |
the new specification for the work. |
| Returns | |
|---|---|
ListenableFuture<WorkManager.UpdateResult> |
a |
Public properties
configuration
abstract val configuration: Configuration
The Configuration instance that WorkManager was initialized with.