A bunch of observables and operators for RxJS.
I created this package as a place to put additional RxJS observables, operators and methods. If you are looking for something that's not in the RxJS distribution, there might be something suitable in here - if you're lucky.
Install the package using NPM:
npm install rxjs-etc --save
Observable
factoriescombineLatestArray, concatArray, forkJoinArray, mergeArray, zipArray
A bunch of static methods that behave in a predictable manner when passed empty arrays. Some of these are now redundant, but some aren't.
To see how these methods behave, consult their tests.
combineLatestHigherOrderArray, combineLatestHigherOrderObject
Higher-order variants of combineLatestArray
- that takes Observable<Observable<T>[]>
and returns Observable<T[]>
- and combineLatestObject
.
combineLatestObject, forkJoinObject, zipObject
Like the array versions, but these take objects. Observable properties are combined using either combineLatest
, forkJoin
or zip
.
Like forkJoin
but only runs the specified number of observables concurrently.
Higher-order variant of mergeArray
- that takes Observable<Observable<T>[]>
and returns Observable<T>
.
Splits a notifier into two or more states and between which notifications are toggled.
Based on expand
. Traverses a graph - with backpressure control - using either a notifier or a consumer.
Works like zipArray
, but if some sources complete whilst others continue to emit values, those the complete are 'padded' with the specified padValue
(which defaults to undefined
).
Runs a sequence of observables in order until an observable completes successfully.
pipe
or let
A bunch of functions that can be passed to the let
operator. Use them like this:
source.let(endWith("this is the end"))
They can also be used with pipe
, like this:
source.pipe(endWith("this is the end"))
Uses a hash function to put values from an observable stream into buckets - which are themselves observable streams. See splitBy
.
Buffers the specified number of most-recent values.
Like defaultIfEmpty
, but it takes a default observable instead of a default value.
Like the RxJava concatMapEager
operator. It accepts a concurrency and eagerly subscribes to its inner observables, buffering their values and then emitting them in the concatMap
order.
Mirrors the source, but sends the last received value to a project function and merges the ObservableInput
that it returns.
Debounce the source observable, but only after the notifier emits a value.
Debounces synchronously emitted values from a source.
Debounce the source observable, but don't debounce the first count
notifications - only the subsequent notifications.
Like debounceTime
, but with an additional duration to ensure some notifications are emitted for super-busy streams.
Delays a source's value notifications until a signal is received from a notifier.
Like finalize
, but calls a child subscription's callback for its parent's.
Like startWith
, but for the other end.
Like filter
, but takes a value - rather than a function - and performs a reference equality check.
Applies the specified TypeScript guard to change the source observable's type and perform a runtime check. Emits an error notification if the guard rejects a value.
Emits true
when the source observable completes.
Like map((value, index) => index)
when it's called without a selector. When called with a selector, it's just an alias for map
.
Like finalize
(which is also exported as an alias), but passes the callback the Notification
that effected the teardown, or undefined
if explicitly unsubscribed.
Apply the operator to the source observable, but select only the initial count
notifications - don't select the subsequent notifications.
Like a combination of startWith
and pairwise
, but with more specific typings.
Like pluck
, but it's type-safe and only lets you valid keys. And it returns the appropriate type.
When creating signals from a source observable - for use with operators that take a notifier, like buffer
and window
- the order in which subscriptions are made is important. prioritize
can be used to ensure that the notifier subscribes to the source first.
A rate limiter with pass through when waiting is not necessary.
Can be used with a ConnectableObservable
instead of refCount
. When the reference count drops to zero, it waits the specified duration and then if the reference count is zero, it unsubscribes. If the reference count is incremented within the duration, no unsubscription occurs.
Somewhat like the change that was made to shareReplay
in 5.5.0.beta.4
. When first subscribed to, a subscription is made to the source, but the source is never explicitly unsubscribed from. Unsubscription from the source only occurs if the source completes or errors.
Like refCount
, but performs connections and unsubscriptions on the specified scheduler.
Emits values using the specified scheduler.
Skips the initial, synchronously emitted values from a source.
Splits an observable stream into two streams. Values that satisfy a predicate are fed into the first stream and values that don't are fed into the second. It's a (better) replacement for partition
- which did not multicast the source. See bucketBy
for the general case of splitting a stream into a specific number of 'buckets'.
Like startWith
but only emits the starting value if the source does not emit within the specified duration.
Apply the operator to the source observable, but don't select the first count
notifications - only the subsequent notifications.
Takes the initial, synchronously emitted values from a source and then completes.
Like takeWhile
, but the value that fails the predicate is taken.
Like tap
, but for subscriptions and unsubscriptions instead of notifications.
Like tap
, but it receives a tuple that includes the emitted value and the index.
Throttle the source observable, but only after the notifier emits a value.
Like subscribeOn
, but for unsubscription.
A bunch of utility functions that do what their names suggest:
isNulled
returns true
if a value is null
or undefined
.