Streams

Living Standard — Last Updated

Participate:
GitHub whatwg/streams (new issue, open issues)
IRC: #whatwg on Freenode
Commits:
GitHub whatwg/streams/commits
Snapshot as of this commit
@streamsstandard
Tests:
web-platform-tests streams/ (ongoing work)
Demos:
streams.spec.whatwg.org/demos
Translation (non-normative):
日本語

Abstract

This specification provides APIs for creating, composing, and consuming streams of data that map efficiently to low-level I/O primitives.

1. Introduction

This section is non-normative.

Large swathes of the web platform are built on streaming data: that is, data that is created, processed, and consumed in an incremental fashion, without ever reading all of it into memory. The Streams Standard provides a common set of APIs for creating and interfacing with such streaming data, embodied in readable streams, writable streams, and transform streams.

These APIs have been designed to efficiently map to low-level I/O primitives, including specializations for byte streams where appropriate. They allow easy composition of multiple streams into pipe chains, or can be used directly via readers and writers. Finally, they are designed to automatically provide backpressure and queuing.

This standard provides the base stream primitives which other parts of the web platform can use to expose their streaming data. For example, [FETCH] exposes Response bodies as ReadableStream instances. More generally, the platform is full of streaming abstractions waiting to be expressed as streams: multimedia streams, file streams, inter-global communication, and more benefit from being able to process data incrementally instead of buffering it all into memory and processing it in one go. By providing the foundation for these streams to be exposed to developers, the Streams Standard enables use cases like:

Web developers can also use the APIs described here to create their own streams, with the same APIs as those provided by the platform. Other developers can then transparently compose platform-provided streams with those supplied by libraries. In this way, the APIs described here provide unifying abstraction for all streams, encouraging an ecosystem to grow around these shared and composable interfaces.

2. Model

A chunk is a single piece of data that is written to or read from a stream. It can be of any type; streams can even contain chunks of different types. A chunk will often not be the most atomic unit of data for a given stream; for example a byte stream might contain chunks consisting of 16 KiB Uint8Arrays, instead of single bytes.

2.1. Readable Streams

A readable stream represents a source of data, from which you can read. In other words, data comes out of a readable stream. Concretely, a readable stream is an instance of the ReadableStream class.

Although a readable stream can be created with arbitrary behavior, most readable streams wrap a lower-level I/O source, called the underlying source. There are two types of underlying source: push sources and pull sources.

Push sources push data at you, whether or not you are listening for it. They may also provide a mechanism for pausing and resuming the flow of data. An example push source is a TCP socket, where data is constantly being pushed from the OS level, at a rate that can be controlled by changing the TCP window size.

Pull sources require you to request data from them. The data may be available synchronously, e.g. if it is held by the operating system’s in-memory buffers, or asynchronously, e.g. if it has to be read from disk. An example pull source is a file handle, where you seek to specific locations and read specific amounts.

Readable streams are designed to wrap both types of sources behind a single, unified interface. For web developer–created streams, the implementation details of a source are provided by an object with certain methods and properties that is passed to the ReadableStream() constructor.

Chunks are enqueued into the stream by the stream’s underlying source. They can then be read one at a time via the stream’s public interface, in particular by using a readable stream reader acquired using the stream’s getReader() method.

Code that reads from a readable stream using its public interface is known as a consumer.

Consumers also have the ability to cancel a readable stream, using its cancel() method. This indicates that the consumer has lost interest in the stream, and will immediately close the stream, throw away any queued chunks, and execute any cancellation mechanism of the underlying source.

Consumers can also tee a readable stream using its tee() method. This will lock the stream, making it no longer directly usable; however, it will create two new streams, called branches, which can be consumed independently.

For streams representing bytes, an extended version of the readable stream is provided to handle bytes efficiently, in particular by minimizing copies. The underlying source for such a readable stream is called an underlying byte source. A readable stream whose underlying source is an underlying byte source is sometimes called a readable byte stream. Consumers of a readable byte stream can acquire a BYOB reader using the stream’s getReader() method.

2.2. Writable Streams

A writable stream represents a destination for data, into which you can write. In other words, data goes in to a writable stream. Concretely, a writable stream is an instance of the WritableStream class.

Analogously to readable streams, most writable streams wrap a lower-level I/O sink, called the underlying sink. Writable streams work to abstract away some of the complexity of the underlying sink, by queuing subsequent writes and only delivering them to the underlying sink one by one.

Chunks are written to the stream via its public interface, and are passed one at a time to the stream’s underlying sink. For web developer-created streams, the implementation details of the sink are provided by an object with certain methods that is passed to the WritableStream() constructor.

Code that writes into a writable stream using its public interface is known as a producer.

Producers also have the ability to abort a writable stream, using its abort() method. This indicates that the producer believes something has gone wrong, and that future writes should be discontinued. It puts the stream in an errored state, even without a signal from the underlying sink.

2.3. Transform Streams

A transform stream consists of a pair of streams: a writable stream, known as its writable side, and a readable stream, known as its readable side. In a manner specific to the transform stream in question, writes to the writable side result in new data being made available for reading from the readable side.

Concretely, any object with a writable property and a readable property can serve as a transform stream. However, the standard TransformStream class makes it much easier to create such a pair that is properly entangled. It wraps a transformer, which defines algorithms for the specific transformation to be performed.

An identity transform stream is a type of transform stream which forwards all chunks written to its writable side to its readable side, without any changes. This can be useful in a variety of scenarios. By default, the TransformStream constructor will create an identity transform stream, when no transform method is present on the transformer object.

Some examples of potential transform streams include:

2.4. Pipe Chains and Backpressure

Streams are primarily used by piping them to each other. A readable stream can be piped directly to a writable stream, using its pipeTo() method, or it can be piped through one or more transform streams first, using its pipeThrough() method.

A set of streams piped together in this way is referred to as a pipe chain. In a pipe chain, the original source is the underlying source of the first readable stream in the chain; the ultimate sink is the underlying sink of the final writable stream in the chain.

Once a pipe chain is constructed, it will propagate signals regarding how fast chunks should flow through it. If any step in the chain cannot yet accept chunks, it propagates a signal backwards through the pipe chain, until eventually the original source is told to stop producing chunks so fast. This process of normalizing flow from the original source according to how fast the chain can process chunks is called backpressure.

Concretely, the original source is given the controller.desiredSize (or byteController.desiredSize) value, and can then adjust its rate of data flow accordingly. This value is derived from the writer.desiredSize corresponding to the ultimate sink, which gets updated as the ultimate sink finishes writing chunks. The pipeTo() method used to construct the chain automatically ensures this information propagates back through the pipe chain.

When teeing a readable stream, the backpressure signals from its two branches will aggregate, such that if neither branch is read from, a backpressure signal will be sent to the underlying source of the original stream.

Piping locks the readable and writable streams, preventing them from being manipulated for the duration of the pipe operation. This allows the implementation to perform important optimizations, such as directly shuttling data from the underlying source to the underlying sink while bypassing many of the intermediate queues.

2.5. Internal Queues and Queuing Strategies

Both readable and writable streams maintain internal queues, which they use for similar purposes. In the case of a readable stream, the internal queue contains chunks that have been enqueued by the underlying source, but not yet read by the consumer. In the case of a writable stream, the internal queue contains chunks which have been written to the stream by the producer, but not yet processed and acknowledged by the underlying sink.

A queuing strategy is an object that determines how a stream should signal backpressure based on the state of its internal queue. The queuing strategy assigns a size to each chunk, and compares the total size of all chunks in the queue to a specified number, known as the high water mark. The resulting difference, high water mark minus total size, is used to determine the desired size to fill the stream’s queue.

For readable streams, an underlying source can use this desired size as a backpressure signal, slowing down chunk generation so as to try to keep the desired size above or at zero. For writable streams, a producer can behave similarly, avoiding writes that would cause the desired size to go negative.

Concretely, a queuing strategy for web developer–created streams is given by any JavaScript object with a highWaterMark property. For byte streams the highWaterMark always has units of bytes. For other streams the default unit is chunks, but a size() function can be included in the strategy object which returns the size for a given chunk. This permits the highWaterMark to be specified in arbitrary floating-point units.

A simple example of a queuing strategy would be one that assigns a size of one to each chunk, and has a high water mark of three. This would mean that up to three chunks could be enqueued in a readable stream, or three chunks written to a writable stream, before the streams are considered to be applying backpressure.

In JavaScript, such a strategy could be written manually as { highWaterMark: 3, size() { return 1; }}, or using the built-in CountQueuingStrategy class, as new CountQueuingStrategy({ highWaterMark: 3 }).

2.6. Locking

A readable stream reader, or simply reader, is an object that allows direct reading of chunks from a readable stream. Without a reader, a consumer can only perform high-level operations on the readable stream: canceling the stream, or piping the readable stream to a writable stream. A reader is acquired via the stream’s getReader() method.

A readable byte stream has the ability to vend two types of readers: default readers and BYOB readers. BYOB ("bring your own buffer") readers allow reading into a developer-supplied buffer, thus minimizing copies. A non-byte readable stream can only vend default readers. Default readers are instances of the ReadableStreamDefaultReader class, while BYOB readers are instances of ReadableStreamBYOBReader.

Similarly, a writable stream writer, or simply writer, is an object that allows direct writing of chunks to a writable stream. Without a writer, a producer can only perform the high-level operations of aborting the stream or piping a readable stream to the writable stream. Writers are represented by the WritableStreamDefaultWriter class.

Under the covers, these high-level operations actually use a reader or writer themselves.

A given readable or writable stream only has at most one reader or writer at a time. We say in this case the stream is locked, and that the reader or writer is active. This state can be determined using the readableStream.locked or writableStream.locked properties.

A reader or writer also has the capability to release its lock, which makes it no longer active, and allows further readers or writers to be acquired. This is done via the defaultReader.releaseLock(), byobReader.releaseLock(), or writer.releaseLock() method, as appropriate.

3. Readable Streams

3.1. Using Readable Streams

The simplest way to consume a readable stream is to simply pipe it to a writable stream. This ensures that backpressure is respected, and any errors (either writing or reading) are propagated through the chain:
readableStream.pipeTo(writableStream)
  .then(() => console.log("All data successfully written!"))
  .catch(e => console.error("Something went wrong!", e));
If you simply want to be alerted of each new chunk from a readable stream, you can pipe it to a new writable stream that you custom-create for that purpose:
readableStream.pipeTo(new WritableStream({
  write(chunk) {
    console.log("Chunk received", chunk);
  },
  close() {
    console.log("All data successfully read!");
  },
  abort(e) {
    console.error("Something went wrong!", e);
  }
}));

By returning promises from your write implementation, you can signal backpressure to the readable stream.

Although readable streams will usually be used by piping them to a writable stream, you can also read them directly by acquiring a reader and using its read() method to get successive chunks. For example, this code logs the next chunk in the stream, if available:
const reader = readableStream.getReader();

reader.read().then(
  ({ value, done }) => {
    if (done) {
      console.log("The stream was already closed!");
    } else {
      console.log(value);
    }
  },
  e => console.error("The stream became errored and cannot be read from!", e)
);

This more manual method of reading a stream is mainly useful for library authors building new high-level operations on streams, beyond the provided ones of piping and teeing.

The above example showed using the readable stream’s default reader. If the stream is a readable byte stream, you can also acquire a BYOB reader for it, which allows more precise control over buffer allocation in order to avoid copies. For example, this code reads the first 1024 bytes from the stream into a single memory buffer:
const reader = readableStream.getReader({ mode: "byob" });

let startingAB = new ArrayBuffer(1024);
readInto(startingAB)
  .then(buffer => console.log("The first 1024 bytes:", buffer))
  .catch(e => console.error("Something went wrong!", e));

function readInto(buffer, offset = 0) {
  if (offset === buffer.byteLength) {
    return Promise.resolve(buffer);
  }

  const view = new Uint8Array(buffer, offset, buffer.byteLength - offset);
  return reader.read(view).then(newView => {
    return readInto(newView.buffer, offset + newView.byteLength);
  });
}

An important thing to note here is that the final buffer value is different from the startingAB, but it (and all intermediate buffers) shares the same backing memory allocation. At each step, the buffer is transferred to a new ArrayBuffer object. The newView is a new Uint8Array, with that ArrayBuffer object as its buffer property, the offset that bytes were written to as its byteOffset property, and the number of bytes that were written as its byteLength property.

3.2. Class ReadableStream

The ReadableStream class is a concrete instance of the general readable stream concept. It is adaptable to any chunk type, and maintains an internal queue to keep track of data supplied by the underlying source but not yet read by any consumer.

3.2.1. Class Definition

This section is non-normative.

If one were to write the ReadableStream class in something close to the syntax of [ECMASCRIPT], it would look like

class ReadableStream {
  constructor(underlyingSource = {}, { size, highWaterMark } = {})

  get locked()

  cancel(reason)
  getReader()
  pipeThrough({ writable, readable }, options)
  pipeTo(dest, { preventClose, preventAbort, preventCancel } = {})
  tee()
}

3.2.2. Internal Slots

Instances of ReadableStream are created with the internal slots described in the following table:

Internal Slot Description (non-normative)
[[disturbed]] A boolean flag set to true when the stream has been read from or canceled
[[readableStreamController]] A ReadableStreamDefaultController or ReadableByteStreamController created with the ability to control the state and queue of this stream; also used for the IsReadableStream brand check
[[reader]] A ReadableStreamDefaultReader or ReadableStreamBYOBReader instance, if the stream is locked to a reader, or undefined if it is not
[[state]] A string containing the stream’s current state, used internally; one of "readable", "closed", or "errored"
[[storedError]] A value indicating how the stream failed, to be given as a failure reason or exception when trying to operate on an errored stream

3.2.3. new ReadableStream(underlyingSource = {}, { size, highWaterMark } = {})

The underlyingSource object passed to the constructor can implement any of the following methods to govern how the constructed stream instance behaves:

Both start and pull are given the ability to manipulate the stream’s internal queue and state via the passed controller object. This is an example of the revealing constructor pattern.

If the underlyingSource object contains a property type set to "bytes", this readable stream is a readable byte stream, and can successfully vend BYOB readers. In that case, the passed controller object will be an instance of ReadableByteStreamController. Otherwise, it will be an instance of ReadableStreamDefaultController.

For readable byte streams, underlyingSource can also contain a property autoAllocateChunkSize, which can be set to a positive integer to enable the auto-allocation feature for this stream. In that case, when a consumer uses a default reader, the stream implementation will automatically allocate an ArrayBuffer of the given size, and call the underlying source code as if the consumer was using a BYOB reader. This can cut down on the amount of code needed when writing the underlying source implementation, as can be seen by comparing §8.3 A readable byte stream with an underlying push source (no backpressure support) without auto-allocation to §8.5 A readable byte stream with an underlying pull source with auto-allocation.

The constructor also accepts a second argument containing the queuing strategy object with two properties: a non-negative number highWaterMark, and a function size(chunk). The supplied strategy could be an instance of the built-in CountQueuingStrategy or ByteLengthQueuingStrategy classes, or it could be custom. If no strategy is supplied, the default behavior will be the same as a CountQueuingStrategy with a high water mark of 1. A size(chunk) function is not meaningful for a readable byte stream, as chunks are always measured in bytes. It is not valid to supply size for a byte stream, or to supply one of the built-in strategies.

  1. Perform ! InitializeReadableStream(this).
  2. Let type be ? GetV(underlyingSource, "type").
  3. Let typeString be ? ToString(type).
  4. If typeString is "bytes",
    1. If highWaterMark is undefined, let highWaterMark be 0.
    2. Set highWaterMark to ? ValidateAndNormalizeHighWaterMark(highWaterMark).
    3. If size is not undefined, throw a RangeError exception.
    4. Perform ? SetUpReadableByteStreamControllerFromUnderlyingSource(this, underlyingSource, highWaterMark).
  5. Otherwise, if type is undefined,
    1. If highWaterMark is undefined, let highWaterMark be 1.
    2. Set highWaterMark to ? ValidateAndNormalizeHighWaterMark(highWaterMark).
    3. Let sizeAlgorithm be ? MakeSizeAlgorithmFromSizeFunction(size).
    4. Perform ? SetUpReadableStreamDefaultControllerFromUnderlyingSource(this, underlyingSource, highWaterMark, sizeAlgorithm).
  6. Otherwise, throw a RangeError exception.

3.2.4. Properties of the ReadableStream Prototype

3.2.4.1. get locked
The locked getter returns whether or not the readable stream is locked to a reader.
  1. If ! IsReadableStream(this) is false, throw a TypeError exception.
  2. Return ! IsReadableStreamLocked(this).
3.2.4.2. cancel(reason)
The cancel method cancels the stream, signaling a loss of interest in the stream by a consumer. The supplied reason argument will be given to the underlying source, which might or might not use it.
  1. If ! IsReadableStream(this) is false, return a promise rejected with a TypeError exception.
  2. If ! IsReadableStreamLocked(this) is true, return a promise rejected with a TypeError exception.
  3. Return ! ReadableStreamCancel(this, reason).
3.2.4.3. getReader({ mode } = {})
The getReader method creates a reader of the type specified by the mode option and locks the stream to the new reader. While the stream is locked, no other reader can be acquired until this one is released.

This functionality is especially useful for creating abstractions that desire the ability to consume a stream in its entirety. By getting a reader for the stream, you can ensure nobody else can interleave reads with yours or cancel the stream, which would interfere with your abstraction.

When mode is undefined, the method creates a default reader (an instance of ReadableStreamDefaultReader). The reader provides the ability to directly read individual chunks from the stream via the reader’s read() method.

When mode is "byob", the getReader method creates a BYOB reader (an instance of ReadableStreamBYOBReader). This feature only works on readable byte streams, i.e. streams which were constructed specifically with the ability to handle "bring your own buffer" reading. The reader provides the ability to directly read individual chunks from the stream via the reader’s read() method, into developer-supplied buffers, allowing more precise control over allocation.

  1. If ! IsReadableStream(this) is false, throw a TypeError exception.
  2. If mode is undefined, return ? AcquireReadableStreamDefaultReader(this).
  3. Set mode to ? ToString(mode).
  4. If mode is "byob", return ? AcquireReadableStreamBYOBReader(this).
  5. Throw a RangeError exception.
An example of an abstraction that might benefit from using a reader is a function like the following, which is designed to read an entire readable stream into memory as an array of chunks.
function readAllChunks(readableStream) {
  const reader = readableStream.getReader();
  const chunks = [];

  return pump();

  function pump() {
    return reader.read().then(({ value, done }) => {
      if (done) {
        return chunks;
      }

      chunks.push(value);
      return pump();
    });
  }
}

Note how the first thing it does is obtain a reader, and from then on it uses the reader exclusively. This ensures that no other consumer can interfere with the stream, either by reading chunks or by canceling the stream.

3.2.4.4. pipeThrough({ writable, readable }, options)
The pipeThrough method provides a convenient, chainable way of piping this readable stream through a transform stream (or any other { writable, readable } pair). It simply pipes the stream into the writable side of the supplied pair, and returns the readable side for further use.

Piping a stream will generally lock it for the duration of the pipe, preventing any other consumer from acquiring a reader.

This method is intentionally generic; it does not require that its this value be a ReadableStream object. It also does not require that its writable argument be a WritableStream instance, or that its readable argument be a ReadableStream instance.

  1. If writable is undefined or readable is undefined, throw a TypeError exception indicating that undefined values are not permitted.

    undefined is special-cased here to make it easier to debug errors such as rs.pipeThrough(writable, readable) which otherwise would silently do nothing. pipeThrough() is otherwise agnostic about the values that it passes through.

  2. Let promise be ? Invoke(this, "pipeTo", « writable, options »).
  3. If Type(promise) is Object and promise has a [[PromiseIsHandled]] internal slot, set promise.[[PromiseIsHandled]] to true.
  4. Return readable.
A typical example of constructing pipe chain using pipeThrough(transform, options) would look like
httpResponseBody
  .pipeThrough(decompressorTransform)
  .pipeThrough(ignoreNonImageFilesTransform)
  .pipeTo(mediaGallery);
3.2.4.5. pipeTo(dest, { preventClose, preventAbort, preventCancel } = {})
The pipeTo method pipes this readable stream to a given writable stream. The way in which the piping process behaves under various error conditions can be customized with a number of passed options. It returns a promise that fulfills when the piping process completes successfully, or rejects if any errors were encountered.

Piping a stream will lock it for the duration of the pipe, preventing any other consumer from acquiring a reader.

Errors and closures of the source and destination streams propagate as follows:

  1. If ! IsReadableStream(this) is false, return a promise rejected with a TypeError exception.
  2. If ! IsWritableStream(dest) is false, return a promise rejected with a TypeError exception.
  3. Set preventClose to ! ToBoolean(preventClose), set preventAbort to ! ToBoolean(preventAbort), and set preventCancel to ! ToBoolean(preventCancel).
  4. If ! IsReadableStreamLocked(this) is true, return a promise rejected with a TypeError exception.
  5. If ! IsWritableStreamLocked(dest) is true, return a promise rejected with a TypeError exception.
  6. If ! IsReadableByteStreamController(this.[[readableStreamController]]) is true, let reader be either ! AcquireReadableStreamBYOBReader(this) or ! AcquireReadableStreamDefaultReader(this), at the user agent’s discretion.
  7. Otherwise, let reader be ! AcquireReadableStreamDefaultReader(this).
  8. Let writer be ! AcquireWritableStreamDefaultWriter(dest).
  9. Let shuttingDown be false.
  10. Let promise be a new promise.
  11. In parallel, using reader and writer, read all chunks from this and write them to dest. Due to the locking provided by the reader and writer, the exact manner in which this happens is not observable to author code, and so there is flexibility in how this is done. The following constraints apply regardless of the exact algorithm used:
    • Public API must not be used: while reading or writing, or performing any of the operations below, the JavaScript-modifiable reader, writer, and stream APIs (i.e. methods on the appropriate prototypes) must not be used. Instead, the streams must be manipulated directly.
    • Backpressure must be enforced:
      • While WritableStreamDefaultWriterGetDesiredSize(writer) is ≤ 0 or is null, the user agent must not read from reader.
      • If reader is a BYOB reader, WritableStreamDefaultWriterGetDesiredSize(writer) should be used to determine the size of the chunks read from reader.
      • Reads or writes should not be delayed for reasons other than these backpressure signals.

        An implementation that waits for each write to successfully complete before proceeding to the next read/write operation violates this recommendation. In doing so, such an implementation makes the internal queue of dest useless, as it ensures dest always contains at most one queued chunk.

    • Shutdown must stop activity: if shuttingDown becomes true, the user agent must not initiate further reads from reader, and must only perform writes of already-read chunks, as described below. In particular, the user agent must check the below conditions before performing any reads or writes, since they might lead to immediate shutdown.
    • Error and close states must be propagated: the following conditions must be applied in order.
      1. Errors must be propagated forward: if this.[[state]] is or becomes "errored", then
        1. If preventAbort is false, shutdown with an action of ! WritableStreamAbort(dest, this.[[storedError]]) and with this.[[storedError]].
        2. Otherwise, shutdown with this.[[storedError]].
      2. Errors must be propagated backward: if dest.[[state]] is or becomes "errored", then
        1. If preventCancel is false, shutdown with an action of ! ReadableStreamCancel(this, dest.[[storedError]]) and with dest.[[storedError]].
        2. Otherwise, shutdown with dest.[[storedError]].
      3. Closing must be propagated forward: if this.[[state]] is or becomes "closed", then
        1. If preventClose is false, shutdown with an action of ! WritableStreamDefaultWriterCloseWithErrorPropagation(writer).
        2. Otherwise, shutdown.
      4. Closing must be propagated backward: if ! WritableStreamCloseQueuedOrInFlight(dest) is true or dest.[[state]] is "closed", then
        1. Assert: no chunks have been read or written.
        2. Let destClosed be a new TypeError.
        3. If preventCancel is false, shutdown with an action of ! ReadableStreamCancel(this, destClosed) and with destClosed.
        4. Otherwise, shutdown with destClosed.
    • Shutdown with an action: if any of the above requirements ask to shutdown with an action action, optionally with an error originalError, then:
      1. If dest.[[state]] is "writable" and ! WritableStreamCloseQueuedOrInFlight(dest) is false,
        1. If any chunks have been read but not yet written, write them to dest.
        2. Wait until every chunk that has been read has been written (i.e. the corresponding promises have settled).
      2. If shuttingDown is true, abort these substeps.
      3. Set shuttingDown to true.
      4. Let p be the result of performing action.
      5. Upon fulfillment of p, finalize, passing along originalError if it was given.
      6. Upon rejection of p with reason newError, finalize with newError.
    • Shutdown: if any of the above requirements or steps ask to shutdown, optionally with an error error, then:
      1. If dest.[[state]] is "writable" and ! WritableStreamCloseQueuedOrInFlight(dest) is false,
        1. If any chunks have been read but not yet written, write them to dest.
        2. Wait until every chunk that has been read has been written (i.e. the corresponding promises have settled).
      2. If shuttingDown is true, abort these substeps.
      3. Set shuttingDown to true.
      4. Finalize, passing along error if it was given.
    • Finalize: both forms of shutdown will eventually ask to finalize, optionally with an error error, which means to perform the following steps:
      1. Perform ! WritableStreamDefaultWriterRelease(writer).
      2. Perform ! ReadableStreamReaderGenericRelease(reader).
      3. If error was given, reject promise with error.
      4. Otherwise, resolve promise with undefined.
  12. Return promise.
3.2.4.6. tee()
The tee method tees this readable stream, returning a two-element array containing the two resulting branches as new ReadableStream instances.

Teeing a stream will lock it, preventing any other consumer from acquiring a reader. To cancel the stream, cancel both of the resulting branches; a composite cancellation reason will then be propagated to the stream’s underlying source.

Note that the chunks seen in each branch will be the same object. If the chunks are not immutable, this could allow interference between the two branches.

  1. If ! IsReadableStream(this) is false, throw a TypeError exception.
  2. Let branches be ? ReadableStreamTee(this, false).
  3. Return ! CreateArrayFromList(branches).
Teeing a stream is most useful when you wish to let two independent consumers read from the stream in parallel, perhaps even at different speeds. For example, given a writable stream cacheEntry representing an on-disk file, and another writable stream httpRequestBody representing an upload to a remote server, you could pipe the same readable stream to both destinations at once:
const [forLocal, forRemote] = readableStream.tee();

Promise.all([
  forLocal.pipeTo(cacheEntry),
  forRemote.pipeTo(httpRequestBody)
])
.then(() => console.log("Saved the stream to the cache and also uploaded it!"))
.catch(e => console.error("Either caching or uploading failed: ", e));

3.3. General Readable Stream Abstract Operations

The following abstract operations, unlike most in this specification, are meant to be generally useful by other specifications, instead of just being part of the implementation of this spec’s classes.

3.3.1. AcquireReadableStreamBYOBReader ( stream ) throws

This abstract operation is meant to be called from other specifications that may wish to acquire a BYOB reader for a given stream.

  1. Return ? Construct(ReadableStreamBYOBReader, « stream »).

3.3.2. AcquireReadableStreamDefaultReader ( stream ) throws

This abstract operation is meant to be called from other specifications that may wish to acquire a default reader for a given stream.

  1. Return ? Construct(ReadableStreamDefaultReader, « stream »).

3.3.3. CreateReadableStream ( startAlgorithm, pullAlgorithm, cancelAlgorithm [, highWaterMark [, sizeAlgorithm ] ] ) throws

This abstract operation is meant to be called from other specifications that wish to create ReadableStream instances. The pullAlgorithm and cancelAlgorithm algorithms must return promises; if supplied, sizeAlgorithm must be an algorithm accepting chunk objects and returning a number; and if supplied, highWaterMark must be a nonnegative, non-NaN number.

CreateReadableStream throws an exception if and only if the supplied startAlgorithm throws.

  1. If highWaterMark was not passed, set it to 1.
  2. If sizeAlgorithm was not passed, set it to an algorithm that returns 1.
  3. Assert: ! IsNonNegativeNumber(highWaterMark) is true.
  4. Let stream be ObjectCreate(the original value of ReadableStream's prototype property).
  5. Perform ! InitializeReadableStream(stream).
  6. Let controller be ObjectCreate(the original value of ReadableStreamDefaultController's prototype property).
  7. Perform ? SetUpReadableStreamDefaultController(stream, controller, startAlgorithm, pullAlgorithm, cancelAlgorithm, highWaterMark, sizeAlgorithm).
  8. Return stream.

3.3.4. CreateReadableByteStream ( startAlgorithm, pullAlgorithm, cancelAlgorithm [, highWaterMark [, autoAllocateChunkSize ] ] ) throws

This abstract operation is meant to be called from other specifications that wish to create ReadableStream instances of type "bytes". The pullAlgorithm and cancelAlgorithm algorithms must return promises; if supplied, highWaterMark must be a nonnegative, non-NaN number, and if supplied, autoAllocateChunkSize must be a positive integer.

CreateReadableByteStream throws an exception if and only if the supplied startAlgorithm throws.

  1. If highWaterMark was not passed, set it to 0.
  2. If autoAllocateChunkSize was not passed, set it to undefined.
  3. Assert: ! IsNonNegativeNumber(highWaterMark) is true.
  4. If autoAllocateChunkSize is not undefined,
    1. Assert: ! IsInteger(autoAllocateChunkSize) is true.
    2. Assert: autoAllocateChunkSize is positive.
  5. Let stream be ObjectCreate(the original value of ReadableStream's prototype property).
  6. Perform ! InitializeReadableStream(stream).
  7. Let controller be ObjectCreate(the original value of ReadableByteStreamController's prototype property).
  8. Perform ? SetUpReadableByteStreamController(stream, controller, startAlgorithm, pullAlgorithm, cancelAlgorithm, highWaterMark, autoAllocateChunkSize).
  9. Return stream.

3.3.5. InitializeReadableStream ( stream ) throws

  1. Set stream.[[state]] to "readable".
  2. Set stream.[[reader]] and stream.[[storedError]] to undefined.
  3. Set stream.[[disturbed]] to false.

3.3.6. IsReadableStream ( x ) nothrow

  1. If Type(x) is not Object, return false.
  2. If x does not have a [[readableStreamController]] internal slot, return false.
  3. Return true.

3.3.7. IsReadableStreamDisturbed ( stream ) nothrow

This abstract operation is meant to be called from other specifications that may wish to query whether or not a readable stream has ever been read from or canceled.

  1. Assert: ! IsReadableStream(stream) is true.
  2. Return stream.[[disturbed]].

3.3.8. IsReadableStreamLocked ( stream ) nothrow

This abstract operation is meant to be called from other specifications that may wish to query whether or not a readable stream is locked to a reader.

  1. Assert: ! IsReadableStream(stream) is true.
  2. If stream.[[reader]] is undefined, return false.
  3. Return true.

3.3.9. ReadableStreamTee ( stream, cloneForBranch2 ) throws

This abstract operation is meant to be called from other specifications that may wish to tee a given readable stream.

The second argument, cloneForBranch2, governs whether or not the data from the original stream will be cloned (using HTML’s serializable objects framework) before appearing in the second of the returned branches. This is useful for scenarios where both branches are to be consumed in such a way that they might otherwise interfere with each other, such as by transferring their chunks. However, it does introduce a noticable asymmetry between the two branches, and limits the possible chunks to serializable ones. [HTML]

In this standard ReadableStreamTee is always called with cloneForBranch2 set to false; other specifications pass true.

  1. Assert: ! IsReadableStream(stream) is true.
  2. Assert: Type(cloneForBranch2) is Boolean.
  3. Let reader be ? AcquireReadableStreamDefaultReader(stream).
  4. Let closedOrErrored be false.
  5. Let canceled1 be false.
  6. Let canceled2 be false.
  7. Let reason1 be undefined.
  8. Let reason2 be undefined.
  9. Let branch1 be undefined.
  10. Let branch2 be undefined.
  11. Let cancelPromise be a new promise.
  12. Let pullAlgorithm be the following steps:
    1. Return the result of transforming ! ReadableStreamDefaultReaderRead(reader) with a fulfillment handler which takes the argument result and performs the following steps:
      1. Assert: Type(result) is Object.
      2. Let value be ? Get(result, "value").
      3. Let done be ? Get(result, "done").
      4. Assert: Type(done) is Boolean.
      5. If done is true and closedOrErrored is false,
        1. If canceled1 is false,
          1. Perform ! ReadableStreamDefaultControllerClose(branch1.[[readableStreamController]]).
        2. If canceled2 is false,
          1. Perform ! ReadableStreamDefaultControllerClose(branch2.[[readableStreamController]]).
        3. Set closedOrErrored to true.
      6. If closedOrErrored is true, return.
      7. Let value1 and value2 be value.
      8. If canceled2 is false and cloneForBranch2 is true, set value2 to ? StructuredDeserialize(? StructuredSerialize(value2), the current Realm Record).
      9. If canceled1 is false, perform ? ReadableStreamDefaultControllerEnqueue(branch1.[[readableStreamController]], value1).
      10. If canceled2 is false, perform ? ReadableStreamDefaultControllerEnqueue(branch2.[[readableStreamController]], value2).
  13. Let cancel1Algorithm be the following steps, taking a reason argument:
    1. Set canceled1 to true.
    2. Set reason1 to reason.
    3. If canceled2 is true,
      1. Let compositeReason be ! CreateArrayFromListreason1, reason2 »).
      2. Let cancelResult be ! ReadableStreamCancel(stream, compositeReason).
      3. Resolve cancelPromise with cancelResult.
    4. Return cancelPromise.
  14. Let cancel2Algorithm be the following steps, taking a reason argument:
    1. Set canceled2 to true.
    2. Set reason2 to reason.
    3. If canceled1 is true,
      1. Let compositeReason be ! CreateArrayFromListreason1, reason2 »).
      2. Let cancelResult be ! ReadableStreamCancel(stream, compositeReason).
      3. Resolve cancelPromise with cancelResult.
    4. Return cancelPromise.
  15. Let startAlgorithm be an algorithm that returns undefined.
  16. Set branch1 to CreateReadableStream(startAlgorithm, pullAlgorithm, cancel1Algorithm).
  17. Set branch2 to CreateReadableStream(startAlgorithm, pullAlgorithm, cancel2Algorithm).
  18. Upon rejection of reader.[[closedPromise]] with reason r,
    1. If closedOrErrored is false, then:
      1. Perform ! ReadableStreamDefaultControllerErrorIfNeeded(branch1.[[readableStreamController]], r).
      2. Perform ! ReadableStreamDefaultControllerErrorIfNeeded(branch2.[[readableStreamController]], r).
      3. Set closedOrErrored to true.
  19. Return « branch1, branch2 ».

3.4. The Interface Between Readable Streams and Controllers

In terms of specification factoring, the way that the ReadableStream class encapsulates the behavior of both simple readable streams and readable byte streams into a single class is by centralizing most of the potentially-varying logic inside the two controller classes, ReadableStreamDefaultController and ReadableByteStreamController. Those classes define most of the stateful internal slots and abstract operations for how a stream’s internal queue is managed and how it interfaces with its underlying source or underlying byte source.

Each controller class defines two internal methods, which are called by the ReadableStream algorithms:

[[CancelSteps]](reason)
The controller’s steps that run in reaction to the stream being canceled, used to clean up the state stored in the controller and inform the underlying source.
[[PullSteps]]()
The controller’s steps that run when a default reader is read from, used to pull from the controller any queued chunks, or pull from the underlying source to get more chunks.

(These are defined as internal methods, instead of as abstract operations, so that they can be called polymorphically by the ReadableStream algorithms, without having to branch on which type of controller is present.)

The rest of this section concerns abstract operations that go in the other direction: they are used by the controller implementations to affect their associated ReadableStream object. This translates internal state changes of the controller into developer-facing results visible through the ReadableStream's public API.

3.4.1. ReadableStreamAddReadIntoRequest ( stream ) nothrow

  1. Assert: ! IsReadableStreamBYOBReader(stream.[[reader]]) is true.
  2. Assert: stream.[[state]] is "readable" or "closed".
  3. Let promise be a new promise.
  4. Let readIntoRequest be Record {[[promise]]: promise}.
  5. Append readIntoRequest as the last element of stream.[[reader]].[[readIntoRequests]].
  6. Return promise.

3.4.2. ReadableStreamAddReadRequest ( stream ) nothrow

  1. Assert: ! IsReadableStreamDefaultReader(stream.[[reader]]) is true.
  2. Assert: stream.[[state]] is "readable".
  3. Let promise be a new promise.
  4. Let readRequest be Record {[[promise]]: promise}.
  5. Append readRequest as the last element of stream.[[reader]].[[readRequests]].
  6. Return promise.

3.4.3. ReadableStreamCancel ( stream, reason ) nothrow

  1. Set stream.[[disturbed]] to true.
  2. If stream.[[state]] is "closed", return a promise resolved with undefined.
  3. If stream.[[state]] is "errored", return a promise rejected with stream.[[storedError]].
  4. Perform ! ReadableStreamClose(stream).
  5. Let sourceCancelPromise be ! stream.[[readableStreamController]].[[CancelSteps]](reason).
  6. Return the result of transforming sourceCancelPromise with a fulfillment handler that returns undefined.

3.4.4. ReadableStreamClose ( stream ) nothrow

  1. Assert: stream.[[state]] is "readable".
  2. Set stream.[[state]] to "closed".
  3. Let reader be stream.[[reader]].
  4. If reader is undefined, return.
  5. If ! IsReadableStreamDefaultReader(reader) is true,
    1. Repeat for each readRequest that is an element of reader.[[readRequests]],
      1. Resolve readRequest.[[promise]] with ! CreateIterResultObject(undefined, true).
    2. Set reader.[[readRequests]] to an empty List.
  6. Resolve reader.[[closedPromise]] with undefined.
The case where stream.[[state]] is "closed", but stream.[[closeRequested]] is false, will happen if the stream was closed without its controller’s close method ever being called: i.e., if the stream was closed by a call to cancel(reason). In this case we allow the controller’s close method to be called and silently do nothing, since the cancelation was outside the control of the underlying source.

3.4.5. ReadableStreamError ( stream, e ) nothrow

  1. Assert: ! IsReadableStream(stream) is true.
  2. Assert: stream.[[state]] is "readable".
  3. Set stream.[[state]] to "errored".
  4. Set stream.[[storedError]] to e.
  5. Let reader be stream.[[reader]].
  6. If reader is undefined, return.
  7. If ! IsReadableStreamDefaultReader(reader) is true,
    1. Repeat for each readRequest that is an element of reader.[[readRequests]],
      1. Reject readRequest.[[promise]] with e.
    2. Set reader.[[readRequests]] to a new empty List.
  8. Otherwise,
    1. Assert: ! IsReadableStreamBYOBReader(reader).
    2. Repeat for each readIntoRequest that is an element of reader.[[readIntoRequests]],
      1. Reject readIntoRequest.[[promise]] with e.
    3. Set reader.[[readIntoRequests]] to a new empty List.
  9. Reject reader.[[closedPromise]] with e.
  10. Set reader.[[closedPromise]].[[PromiseIsHandled]] to true.

3.4.6. ReadableStreamFulfillReadIntoRequest ( stream, chunk, done ) nothrow

  1. Let reader be stream.[[reader]].
  2. Let readIntoRequest be the first element of reader.[[readIntoRequests]].
  3. Remove readIntoRequest from reader.[[readIntoRequests]], shifting all other elements downward (so that the second becomes the first, and so on).
  4. Resolve readIntoRequest.[[promise]] with ! CreateIterResultObject(chunk, done).

3.4.7. ReadableStreamFulfillReadRequest ( stream, chunk, done ) nothrow

  1. Let reader be stream.[[reader]].
  2. Let readRequest be the first element of reader.[[readRequests]].
  3. Remove readRequest from reader.[[readRequests]], shifting all other elements downward (so that the second becomes the first, and so on).
  4. Resolve readRequest.[[promise]] with ! CreateIterResultObject(chunk, done).

3.4.8. ReadableStreamGetNumReadIntoRequests ( stream ) nothrow

  1. Return the number of elements in stream.[[reader]].[[readIntoRequests]].

3.4.9. ReadableStreamGetNumReadRequests ( stream ) nothrow

  1. Return the number of elements in stream.[[reader]].[[readRequests]].

3.4.10. ReadableStreamHasBYOBReader ( stream ) nothrow

  1. Let reader be stream.[[reader]].
  2. If reader is undefined, return false.
  3. If ! IsReadableStreamBYOBReader(reader) is false, return false.
  4. Return true.

3.4.11. ReadableStreamHasDefaultReader ( stream ) nothrow

  1. Let reader be stream.[[reader]].
  2. If reader is undefined, return false.
  3. If ! IsReadableStreamDefaultReader(reader) is false, return false.
  4. Return true.

3.5. Class ReadableStreamDefaultReader

The ReadableStreamDefaultReader class represents a default reader designed to be vended by a ReadableStream instance.

3.5.1. Class Definition

This section is non-normative.

If one were to write the ReadableStreamDefaultReader class in something close to the syntax of [ECMASCRIPT], it would look like

class ReadableStreamDefaultReader {
  constructor(stream)

  get closed()

  cancel(reason)
  read()
  releaseLock()
}

3.5.2. Internal Slots

Instances of ReadableStreamDefaultReader are created with the internal slots described in the following table:

Internal Slot Description (non-normative)
[[closedPromise]] A promise returned by the reader’s closed getter
[[ownerReadableStream]] A ReadableStream instance that owns this reader
[[readRequests]] A List of promises returned by calls to the reader’s read() method that have not yet been resolved, due to the consumer requesting chunks sooner than they are available; also used for the IsReadableStreamDefaultReader brand check

3.5.3. new ReadableStreamDefaultReader(stream)

The ReadableStreamDefaultReader constructor is generally not meant to be used directly; instead, a stream’s getReader() method ought to be be used.
  1. If ! IsReadableStream(stream) is false, throw a TypeError exception.
  2. If ! IsReadableStreamLocked(stream) is true, throw a TypeError exception.
  3. Perform ! ReadableStreamReaderGenericInitialize(this, stream).
  4. Set this.[[readRequests]] to a new empty List.

3.5.4. Properties of the ReadableStreamDefaultReader Prototype

3.5.4.1. get closed
The closed getter returns a promise that will be fulfilled when the stream becomes closed or the reader’s lock is released, or rejected if the stream ever errors.
  1. If ! IsReadableStreamDefaultReader(this) is false, return a promise rejected with a TypeError exception.
  2. Return this.[[closedPromise]].
3.5.4.2. cancel(reason)
If the reader is active, the cancel method behaves the same as that for the associated stream.
  1. If ! IsReadableStreamDefaultReader(this) is false, return a promise rejected with a TypeError exception.
  2. If this.[[ownerReadableStream]] is undefined, return a promise rejected with a TypeError exception.
  3. Return ! ReadableStreamReaderGenericCancel(this, reason).
3.5.4.3. read()
The read method will return a promise that allows access to the next chunk from the stream’s internal queue, if available.

If reading a chunk causes the queue to become empty, more data will be pulled from the underlying source.

  1. If ! IsReadableStreamDefaultReader(this) is false, return a promise rejected with a TypeError exception.
  2. If this.[[ownerReadableStream]] is undefined, return a promise rejected with a TypeError exception.
  3. Return ! ReadableStreamDefaultReaderRead(this).
3.5.4.4. releaseLock()
The releaseLock method releases the reader’s lock on the corresponding stream. After the lock is released, the reader is no longer active. If the associated stream is errored when the lock is released, the reader will appear errored in the same way from now on; otherwise, the reader will appear closed.

A reader’s lock cannot be released while it still has a pending read request, i.e., if a promise returned by the reader’s read() method has not yet been settled. Attempting to do so will throw a TypeError and leave the reader locked to the stream.

  1. If ! IsReadableStreamDefaultReader(this) is false, throw a TypeError exception.
  2. If this.[[ownerReadableStream]] is undefined, return.
  3. If this.[[readRequests]] is not empty, throw a TypeError exception.
  4. Perform ! ReadableStreamReaderGenericRelease(this).

3.6. Class ReadableStreamBYOBReader

The ReadableStreamBYOBReader class represents a BYOB reader designed to be vended by a ReadableStream instance.

3.6.1. Class Definition

This section is non-normative.

If one were to write the ReadableStreamBYOBReader class in something close to the syntax of [ECMASCRIPT], it would look like

class ReadableStreamBYOBReader {
  constructor(stream)

  get closed()

  cancel(reason)
  read(view)
  releaseLock()
}

3.6.2. Internal Slots

Instances of ReadableStreamBYOBReader are created with the internal slots described in the following table:

Internal Slot Description (non-normative)
[[closedPromise]] A promise returned by the reader’s closed getter
[[ownerReadableStream]] A ReadableStream instance that owns this reader
[[readIntoRequests]] A List of promises returned by calls to the reader’s read(view) method that have not yet been resolved, due to the consumer requesting chunks sooner than they are available; also used for the IsReadableStreamBYOBReader brand check

3.6.3. new ReadableStreamBYOBReader(stream)

The ReadableStreamBYOBReader constructor is generally not meant to be used directly; instead, a stream’s getReader() method ought to be used.
  1. If ! IsReadableStream(stream) is false, throw a TypeError exception.
  2. If ! IsReadableByteStreamController(stream.[[readableStreamController]]) is false, throw a TypeError exception.
  3. If ! IsReadableStreamLocked(stream) is true, throw a TypeError exception.
  4. Perform ! ReadableStreamReaderGenericInitialize(this, stream).
  5. Set this.[[readIntoRequests]] to a new empty List.

3.6.4. Properties of the ReadableStreamBYOBReader Prototype

3.6.4.1. get closed
The closed getter returns a promise that will be fulfilled when the stream becomes closed or the reader’s lock is released, or rejected if the stream ever errors.
  1. If ! IsReadableStreamBYOBReader(this) is false, return a promise rejected with a TypeError exception.
  2. Return this.[[closedPromise]].
3.6.4.2. cancel(reason)
If the reader is active, the cancel method behaves the same as that for the associated stream.
  1. If ! IsReadableStreamBYOBReader(this) is false, return a promise rejected with a TypeError exception.
  2. If this.[[ownerReadableStream]] is undefined, return a promise rejected with a TypeError exception.
  3. Return ! ReadableStreamReaderGenericCancel(this, reason).
3.6.4.3. read(view)
The read method will write read bytes into view and return a promise resolved with a possibly transferred buffer as described below.

If reading a chunk causes the queue to become empty, more data will be pulled from the underlying byte source.

  1. If ! IsReadableStreamBYOBReader(this) is false, return a promise rejected with a TypeError exception.
  2. If this.[[ownerReadableStream]] is undefined, return a promise rejected with a TypeError exception.
  3. If Type(view) is not Object, return a promise rejected with a TypeError exception.
  4. If view does not have a [[ViewedArrayBuffer]] internal slot, return a promise rejected with a TypeError exception.
  5. If ! IsDetachedBuffer(view.[[ViewedArrayBuffer]]) is true, return a promise rejected with a TypeError exception.
  6. If view.[[ByteLength]] is 0, return a promise rejected with a TypeError exception.
  7. Return ! ReadableStreamBYOBReaderRead(this, view).
3.6.4.4. releaseLock()
The releaseLock method releases the reader’s lock on the corresponding stream. After the lock is released, the reader is no longer active. If the associated stream is errored when the lock is released, the reader will appear errored in the same way from now on; otherwise, the reader will appear closed.

A reader’s lock cannot be released while it still has a pending read request, i.e., if a promise returned by the reader’s read() method has not yet been settled. Attempting to do so will throw a TypeError and leave the reader locked to the stream.

  1. If ! IsReadableStreamBYOBReader(this) is false, throw a TypeError exception.
  2. If this.[[ownerReadableStream]] is undefined, return.
  3. If this.[[readIntoRequests]] is not empty, throw a TypeError exception.
  4. Perform ! ReadableStreamReaderGenericRelease(this).

3.7. Readable Stream Reader Abstract Operations

3.7.1. IsReadableStreamDefaultReader ( x ) nothrow

  1. If Type(x) is not Object, return false.
  2. If x does not have a [[readRequests]] internal slot, return false.
  3. Return true.

3.7.2. IsReadableStreamBYOBReader ( x ) nothrow

  1. If Type(x) is not Object, return false.
  2. If x does not have a [[readIntoRequests]] internal slot, return false.
  3. Return true.

3.7.3. ReadableStreamReaderGenericCancel ( reader, reason ) nothrow

  1. Let stream be reader.[[ownerReadableStream]].
  2. Assert: stream is not undefined.
  3. Return ! ReadableStreamCancel(stream, reason).

3.7.4. ReadableStreamReaderGenericInitialize ( reader, stream ) nothrow

  1. Set reader.[[ownerReadableStream]] to stream.
  2. Set stream.[[reader]] to reader.
  3. If stream.[[state]] is "readable",
    1. Set reader.[[closedPromise]] to a new promise.
  4. Otherwise, if stream.[[state]] is "closed",
    1. Set reader.[[closedPromise]] to a promise resolved with undefined.
  5. Otherwise,
    1. Assert: stream.[[state]] is "errored".
    2. Set reader.[[closedPromise]] to a promise rejected with stream.[[storedError]].
    3. Set reader.[[closedPromise]].[[PromiseIsHandled]] to true.

3.7.5. ReadableStreamReaderGenericRelease ( reader ) nothrow

  1. Assert: reader.[[ownerReadableStream]] is not undefined.
  2. Assert: reader.[[ownerReadableStream]].[[reader]] is reader.
  3. If reader.[[ownerReadableStream]].[[state]] is "readable", reject reader.[[closedPromise]] with a TypeError exception.
  4. Otherwise, set reader.[[closedPromise]] to a promise rejected with a TypeError exception.
  5. Set reader.[[closedPromise]].[[PromiseIsHandled]] to true.
  6. Set reader.[[ownerReadableStream]].[[reader]] to undefined.
  7. Set reader.[[ownerReadableStream]] to undefined.

3.7.6. ReadableStreamBYOBReaderRead ( reader, view ) nothrow

  1. Let stream be reader.[[ownerReadableStream]].
  2. Assert: stream is not undefined.
  3. Set stream.[[disturbed]] to true.
  4. If stream.[[state]] is "errored", return a promise rejected with stream.[[storedError]].
  5. Return ! ReadableByteStreamControllerPullInto(stream.[[readableStreamController]], view).

3.7.7. ReadableStreamDefaultReaderRead ( reader ) nothrow

  1. Let stream be reader.[[ownerReadableStream]].
  2. Assert: stream is not undefined.
  3. Set stream.[[disturbed]] to true.
  4. If stream.[[state]] is "closed", return a promise resolved with ! CreateIterResultObject(undefined, true).
  5. If stream.[[state]] is "errored", return a promise rejected with stream.[[storedError]].
  6. Assert: stream.[[state]] is "readable".
  7. Return ! stream.[[readableStreamController]].[[PullSteps]]().

3.8. Class ReadableStreamDefaultController

The ReadableStreamDefaultController class has methods that allow control of a ReadableStream's state and internal queue. When constructing a ReadableStream that is not a readable byte stream, the underlying source is given a corresponding ReadableStreamDefaultController instance to manipulate.

3.8.1. Class Definition

This section is non-normative.

If one were to write the ReadableStreamDefaultController class in something close to the syntax of [ECMASCRIPT], it would look like

class ReadableStreamDefaultController {
  constructor() // always throws

  get desiredSize()

  close()
  enqueue(chunk)
  error(e)
}

3.8.2. Internal Slots

Instances of ReadableStreamDefaultController are created with the internal slots described in the following table:

Internal Slot Description (non-normative)
[[cancelAlgorithm]] A promise-returning algorithm, taking one argument (the cancel reason), which communicates a requested cancelation to the underlying source
[[closeRequested]] A boolean flag indicating whether the stream has been closed by its underlying source, but still has chunks in its internal queue that have not yet been read
[[controlledReadableStream]] The ReadableStream instance controlled
[[pullAgain]] A boolean flag set to true if the stream’s mechanisms requested a call to the underlying source’s pull algorithm to pull more data, but the pull could not yet be done since a previous call is still executing
[[pullAlgorithm]] A promise-returning algorithm that pulls data from the underlying source
[[pulling]] A boolean flag set to true while the underlying source’s pull algorithm is executing and the returned promise has not yet fulfilled, used to prevent reentrant calls
[[queue]] A List representing the stream’s internal queue of chunks
[[queueTotalSize]] The total size of all the chunks stored in [[queue]] (see §6.3 Queue-with-Sizes Operations)
[[started]] A boolean flag indicating whether the underlying source has finished starting
[[strategyHWM]] A number supplied to the constructor as part of the stream’s queuing strategy, indicating the point at which the stream will apply backpressure to its underlying source
[[strategySizeAlgorithm]] An algorithm to calculate the size of enqueued chunks, as part of the stream’s queuing strategy

3.8.3. new ReadableStreamDefaultController()

The ReadableStreamDefaultController constructor cannot be used directly; ReadableStreamDefaultController instances are created automatically during ReadableStream construction.
  1. Throw a TypeError.

3.8.4. Properties of the ReadableStreamDefaultController Prototype

3.8.4.1. get desiredSize
The desiredSize getter returns the desired size to fill the controlled stream’s internal queue. It can be negative, if the queue is over-full. An underlying source ought to use this information to determine when and how to apply backpressure.
  1. If ! IsReadableStreamDefaultController(this) is false, throw a TypeError exception.
  2. Return ! ReadableStreamDefaultControllerGetDesiredSize(this).
3.8.4.2. close()
The close method will close the controlled readable stream. Consumers will still be able to read any previously-enqueued chunks from the stream, but once those are read, the stream will become closed.
  1. If ! IsReadableStreamDefaultController(this) is false, throw a TypeError exception.
  2. If ! ReadableStreamDefaultControllerCanCloseOrEnqueue(this) is false, throw a TypeError exception.
  3. Perform ! ReadableStreamDefaultControllerClose(this).
3.8.4.3. enqueue(chunk)
The enqueue method will enqueue a given chunk in the controlled readable stream.
  1. If ! IsReadableStreamDefaultController(this) is false, throw a TypeError exception.
  2. If ! ReadableStreamDefaultControllerCanCloseOrEnqueue(this) is false, throw a TypeError exception.
  3. Return ? ReadableStreamDefaultControllerEnqueue(this, chunk).
3.8.4.4. error(e)
The error method will error the readable stream, making all future interactions with it fail with the given error e.
  1. If ! IsReadableStreamDefaultController(this) is false, throw a TypeError exception.
  2. Let stream be this.[[controlledReadableStream]].
  3. If stream.[[state]] is not "readable", throw a TypeError exception.
  4. Perform ! ReadableStreamDefaultControllerError(this, e).

3.8.5. Readable Stream Default Controller Internal Methods

The following are additional internal methods implemented by each ReadableStreamDefaultController instance. The readable stream implementation will polymorphically call to either these or their counterparts for BYOB controllers.

3.8.5.1. [[CancelSteps]](reason)
  1. Perform ! ResetQueue(this).
  2. Return the result of performing this.[[cancelAlgorithm]], passing reason.
3.8.5.2. [[PullSteps]]()
  1. Let stream be this.[[controlledReadableStream]].
  2. If this.[[queue]] is not empty,
    1. Let chunk be ! DequeueValue(this).
    2. If this.[[closeRequested]] is true and this.[[queue]] is empty, perform ! ReadableStreamClose(stream).
    3. Otherwise, perform ! ReadableStreamDefaultControllerCallPullIfNeeded(this).
    4. Return a promise resolved with ! CreateIterResultObject(chunk, false).
  3. Let pendingPromise be ! ReadableStreamAddReadRequest(stream).
  4. Perform ! ReadableStreamDefaultControllerCallPullIfNeeded(this).
  5. Return pendingPromise.

3.9. Readable Stream Default Controller Abstract Operations

3.9.1. IsReadableStreamDefaultController ( x ) nothrow

  1. If Type(x) is not Object, return false.
  2. If x does not have a [[controlledReadableStream]] internal slot, return false.
  3. Return true.

3.9.2. ReadableStreamDefaultControllerCallPullIfNeeded ( controller ) nothrow

  1. Let shouldPull be ! ReadableStreamDefaultControllerShouldCallPull(controller).
  2. If shouldPull is false, return.
  3. If controller.[[pulling]] is true,
    1. Set controller.[[pullAgain]] to true.
    2. Return.
  4. Assert: controller.[[pullAgain]] is false.
  5. Set controller.[[pulling]] to true.
  6. Let pullPromise be the result of performing controller.[[pullAlgorithm]].
  7. Upon fulfillment of pullPromise,
    1. Set controller.[[pulling]] to false.
    2. If controller.[[pullAgain]] is true,
      1. Set controller.[[pullAgain]] to false.
      2. Perform ! ReadableStreamDefaultControllerCallPullIfNeeded(controller).
  8. Upon rejection of pullPromise with reason e,
    1. Perform ! ReadableStreamDefaultControllerErrorIfNeeded(controller, e).

3.9.3. ReadableStreamDefaultControllerShouldCallPull ( controller ) nothrow

  1. Let stream be controller.[[controlledReadableStream]].
  2. If ! ReadableStreamDefaultControllerCanCloseOrEnqueue(controller) is false, return false.
  3. If controller.[[started]] is false, return false.
  4. If ! IsReadableStreamLocked(stream) is true and ! ReadableStreamGetNumReadRequests(stream) > 0, return true.
  5. Let desiredSize be ReadableStreamDefaultControllerGetDesiredSize(controller).
  6. If desiredSize > 0, return true.
  7. Return false.

3.9.4. ReadableStreamDefaultControllerClose ( controller ) nothrow

This abstract operation can be called by other specifications that wish to close a readable stream, in the same way a developer-created stream would be closed by its associated controller object. Specifications should not do this to streams they did not create, and must ensure they have obeyed the preconditions (listed here as asserts).

  1. Let stream be controller.[[controlledReadableStream]].
  2. Assert: ! ReadableStreamDefaultControllerCanCloseOrEnqueue(controller) is true.
  3. Set controller.[[closeRequested]] to true.
  4. If controller.[[queue]] is empty, perform ! ReadableStreamClose(stream).

3.9.5. ReadableStreamDefaultControllerEnqueue ( controller, chunk ) throws

This abstract operation can be called by other specifications that wish to enqueue chunks in a readable stream, in the same way a developer would enqueue chunks using the stream’s associated controller object. Specifications should not do this to streams they did not create, and must ensure they have obeyed the preconditions (listed here as asserts).

  1. Let stream be controller.[[controlledReadableStream]].
  2. Assert: ! ReadableStreamDefaultControllerCanCloseOrEnqueue(controller) is true.
  3. If ! IsReadableStreamLocked(stream) is true and ! ReadableStreamGetNumReadRequests(stream) > 0, perform ! ReadableStreamFulfillReadRequest(stream, chunk, false).
  4. Otherwise,
    1. Let result be the result of performing controller.[[strategySizeAlgorithm]], passing in chunk, and interpreting the result as an ECMAScript completion value.
    2. If result is an abrupt completion,
      1. Perform ! ReadableStreamDefaultControllerErrorIfNeeded(controller, result.[[Value]]).
      2. Return result.
    3. Let chunkSize be result.[[Value]].
    4. Let enqueueResult be EnqueueValueWithSize(controller, chunk, chunkSize).
    5. If enqueueResult is an abrupt completion,
      1. Perform ! ReadableStreamDefaultControllerErrorIfNeeded(controller, enqueueResult.[[Value]]).
      2. Return enqueueResult.
  5. Perform ! ReadableStreamDefaultControllerCallPullIfNeeded(controller).

3.9.6. ReadableStreamDefaultControllerError ( controller, e ) nothrow

This abstract operation can be called by other specifications that wish to move a readable stream to an errored state, in the same way a developer would error a stream using its associated controller object. Specifications should not do this to streams they did not create, and must ensure they have obeyed the precondition (listed here as an assert).

  1. Let stream be controller.[[controlledReadableStream]].
  2. Assert: stream.[[state]] is "readable".
  3. Perform ! ResetQueue(controller).
  4. Perform ! ReadableStreamError(stream, e).

3.9.7. ReadableStreamDefaultControllerErrorIfNeeded ( controller, e ) nothrow

  1. If controller.[[controlledReadableStream]].[[state]] is "readable", perform ! ReadableStreamDefaultControllerError(controller, e).

3.9.8. ReadableStreamDefaultControllerGetDesiredSize ( controller ) nothrow

This abstract operation can be called by other specifications that wish to determine the desired size to fill this stream’s internal queue, similar to how a developer would consult the desiredSize property of the stream’s associated controller object. Specifications should not use this on streams they did not create.

  1. Let stream be controller.[[controlledReadableStream]].
  2. Let state be stream.[[state]].
  3. If state is "errored", return null.
  4. If state is "closed", return 0.
  5. Return controller.[[strategyHWM]] − controller.[[queueTotalSize]].

3.9.9. ReadableStreamDefaultControllerHasBackpressure ( controller ) nothrow

This abstract operation is used in the implementation of TransformStream.

  1. If ! ReadableStreamDefaultControllerShouldCallPull(controller) is true, return false.
  2. Otherwise, return true.

3.9.10. ReadableStreamDefaultControllerCanCloseOrEnqueue ( controller ) nothrow

  1. Let state be controller.[[controlledReadableStream]].[[state]].
  2. If controller.[[closeRequested]] is false and state is "readable", return true.
  3. Otherwise, return false.
The case where stream.[[closeRequested]] is false, but stream.[[state]] is not "readable", happens when the stream is errored via error(e), or when it is closed without its controller’s close method ever being called: e.g., if the stream was closed by a call to cancel(reason).

3.9.11. SetUpReadableStreamDefaultController(stream, controller, startAlgorithm, pullAlgorithm, cancelAlgorithm, highWaterMark, sizeAlgorithm ) throws

  1. Assert: stream.[[readableStreamController]] is undefined.
  2. Set controller.[[controlledReadableStream]] to stream.
  3. Set controller.[[queue]] and controller.[[queueSize]] to undefined, then perform ! ResetQueue(controller).
  4. Set controller.[[started]], controller.[[closeRequested]], controller.[[pullAgain]], and controller.[[pulling]] to false.
  5. Set controller.[[strategySizeAlgorithm]] to sizeAlgorithm and controller.[[strategyHWM]] to highWaterMark.
  6. Set controller.[[pullAlgorithm]] to pullAlgorithm.
  7. Set controller.[[cancelAlgorithm]] to cancelAlgorithm.
  8. Set stream.[[readableStreamController]] to controller.
  9. Let startResult be the result of performing startAlgorithm. (This may throw an exception.)
  10. Let startPromise be a promise resolved with startResult.
  11. Upon fulfillment of startPromise,
    1. Set controller.[[started]] to true.
    2. Assert: controller.[[pulling]] is false.
    3. Assert: controller.[[pullAgain]] is false.
    4. Perform ! ReadableStreamDefaultControllerCallPullIfNeeded(controller).
  12. Upon rejection of startPromise with reason r,
    1. Perform ! ReadableStreamDefaultControllerErrorIfNeeded(controller, r).

3.9.12. SetUpReadableStreamDefaultControllerFromUnderlyingSource(stream, underlyingSource, highWaterMark, sizeAlgorithm ) throws

  1. Assert: underlyingSource is not undefined.
  2. Let controller be ObjectCreate(the original value of ReadableStreamDefaultController's prototype property).
  3. Let startAlgorithm be the following steps:
    1. Return ? InvokeOrNoop(underlyingSource, "start", « controller »).
  4. Let pullAlgorithm be ? CreateAlgorithmFromUnderlyingMethod(underlyingSource, "pull", 0, « controller »).
  5. Let cancelAlgorithm be ? CreateAlgorithmFromUnderlyingMethod(underlyingSource, "cancel", 1, « »).
  6. Perform ? SetUpReadableStreamDefaultController(stream, controller, startAlgorithm, pullAlgorithm, cancelAlgorithm, highWaterMark, sizeAlgorithm).

3.10. Class ReadableByteStreamController

The ReadableByteStreamController class has methods that allow control of a ReadableStream's state and internal queue. When constructing a ReadableStream, the underlying byte source is given a corresponding ReadableByteStreamController instance to manipulate.

3.10.1. Class Definition

This section is non-normative.

If one were to write the ReadableByteStreamController class in something close to the syntax of [ECMASCRIPT], it would look like

class ReadableByteStreamController {
  constructor() // always throws

  get byobRequest()
  get desiredSize()

  close()
  enqueue(chunk)
  error(e)
}

3.10.2. Internal Slots

Instances of ReadableByteStreamController are created with the internal slots described in the following table:

Internal Slot Description (non-normative)
[[autoAllocateChunkSize]] A positive integer, when the automatic buffer allocation feature is enabled. In that case, this value specifies the size of buffer to allocate. It is undefined otherwise.
[[byobRequest]] A ReadableStreamBYOBRequest instance representing the current BYOB pull request
[[cancelAlgorithm]] A promise-returning algorithm, taking one argument (the cancel reason), which communicates a requested cancelation to the underlying source
[[closeRequested]] A boolean flag indicating whether the stream has been closed by its underlying byte source, but still has chunks in its internal queue that have not yet been read
[[controlledReadableByteStream]] The ReadableStream instance controlled
[[pullAgain]] A boolean flag set to true if the stream’s mechanisms requested a call to the underlying byte source’s pull method to pull more data, but the pull could not yet be done since a previous call is still executing
[[pullAlgorithm]] A promise-returning algorithm that pulls data from the underlying source
[[pulling]] A boolean flag set to true while the underlying byte source’s pull method is executing and has not yet fulfilled, used to prevent reentrant calls
[[pendingPullIntos]] A List of descriptors representing pending BYOB pull requests
[[queue]] A List representing the stream’s internal queue of chunks
[[queueTotalSize]] The total size (in bytes) of all the chunks stored in [[queue]]
[[started]] A boolean flag indicating whether the underlying source has finished starting
[[strategyHWM]] A number supplied to the constructor as part of the stream’s queuing strategy, indicating the point at which the stream will apply backpressure to its underlying byte source

Although ReadableByteStreamController instances have [[queue]] and [[queueTotalSize]] slots, we do not use most of the abstract operations in §6.3 Queue-with-Sizes Operations on them, as the way in which we manipulate this queue is rather different than the others in the spec. Instead, we update the two slots together manually.

This might be cleaned up in a future spec refactoring.

3.10.3. new ReadableByteStreamController()

The ReadableByteStreamController constructor cannot be used directly; ReadableByteStreamController instances are created automatically during ReadableStream construction.
  1. Throw a TypeError exception.

3.10.4. Properties of the ReadableByteStreamController Prototype

3.10.4.1. get byobRequest
The byobRequest getter returns the current BYOB pull request.
  1. If IsReadableByteStreamController(this) is false, throw a TypeError exception.
  2. If this.[[byobRequest]] is undefined and this.[[pendingPullIntos]] is not empty,
    1. Let firstDescriptor be the first element of this.[[pendingPullIntos]].
    2. Let view be ! Construct(%Uint8Array%, « firstDescriptor.[[buffer]], firstDescriptor.[[byteOffset]] + firstDescriptor.[[bytesFilled]], firstDescriptor.[[byteLength]] − firstDescriptor.[[bytesFilled]] »).
    3. Set this.[[byobRequest]] to ! Construct(ReadableStreamBYOBRequest, « this, view »).
  3. Return this.[[byobRequest]].
3.10.4.2. get desiredSize
The desiredSize getter returns the desired size to fill the controlled stream’s internal queue. It can be negative, if the queue is over-full. An underlying source ought to use this information to determine when and how to apply backpressure.
  1. If ! IsReadableByteStreamController(this) is false, throw a TypeError exception.
  2. Return ! ReadableByteStreamControllerGetDesiredSize(this).
3.10.4.3. close()
The close method will close the controlled readable stream. Consumers will still be able to read any previously-enqueued chunks from the stream, but once those are read, the stream will become closed.
  1. If ! IsReadableByteStreamController(this) is false, throw a TypeError exception.
  2. If this.[[closeRequested]] is true, throw a TypeError exception.
  3. If this.[[controlledReadableByteStream]].[[state]] is not "readable", throw a TypeError exception.
  4. Perform ? ReadableByteStreamControllerClose(this).
3.10.4.4. enqueue(chunk)
The enqueue method will enqueue a given chunk in the controlled readable stream.
  1. If ! IsReadableByteStreamController(this) is false, throw a TypeError exception.
  2. If this.[[closeRequested]] is true, throw a TypeError exception.
  3. If this.[[controlledReadableByteStream]].[[state]] is not "readable", throw a TypeError exception.
  4. If Type(chunk) is not Object, throw a TypeError exception.
  5. If chunk does not have a [[ViewedArrayBuffer]] internal slot, throw a TypeError exception.
  6. If ! IsDetachedBuffer(chunk.[[ViewedArrayBuffer]]) is true, throw a TypeError exception.
  7. Return ! ReadableByteStreamControllerEnqueue(this, chunk).
3.10.4.5. error(e)
The error method will error the readable stream, making all future interactions with it fail with the given error e.
  1. If ! IsReadableByteStreamController(this) is false, throw a TypeError exception.
  2. Let stream be this.[[controlledReadableByteStream]].
  3. If stream.[[state]] is not "readable", throw a TypeError exception.
  4. Perform ! ReadableByteStreamControllerError(this, e).

3.10.5. Readable Stream BYOB Controller Internal Methods

The following are additional internal methods implemented by each ReadableByteStreamController instance. The readable stream implementation will polymorphically call to either these or their counterparts for default controllers.

3.10.5.1. [[CancelSteps]](reason)
  1. If this.[[pendingPullIntos]] is not empty,
    1. Let firstDescriptor be the first element of this.[[pendingPullIntos]].
    2. Set firstDescriptor.[[bytesFilled]] to 0.
  2. Perform ! ResetQueue(this).
  3. Return the result of performing this.[[cancelAlgorithm]], passing in reason.
3.10.5.2. [[PullSteps]]()
  1. Let stream be this.[[controlledReadableByteStream]].
  2. Assert: ! ReadableStreamHasDefaultReader(stream) is true.
  3. If this.[[queueTotalSize]] > 0,
    1. Assert: ! ReadableStreamGetNumReadRequests(stream) is 0.
    2. Let entry be the first element of this.[[queue]].
    3. Remove entry from this.[[queue]], shifting all other elements downward (so that the second becomes the first, and so on).
    4. Set this.[[queueTotalSize]] to this.[[queueTotalSize]] − entry.[[byteLength]].
    5. Perform ! ReadableByteStreamControllerHandleQueueDrain(this).
    6. Let view be ! Construct(%Uint8Array%, « entry.[[buffer]], entry.[[byteOffset]], entry.[[byteLength]] »).
    7. Return a promise resolved with ! CreateIterResultObject(view, false).
  4. Let autoAllocateChunkSize be this.[[autoAllocateChunkSize]].
  5. If autoAllocateChunkSize is not undefined,
    1. Let buffer be Construct(%ArrayBuffer%, « autoAllocateChunkSize »).
    2. If buffer is an abrupt completion, return a promise rejected with buffer.[[Value]].
    3. Let pullIntoDescriptor be Record {[[buffer]]: buffer.[[Value]], [[byteOffset]]: 0, [[byteLength]]: autoAllocateChunkSize, [[bytesFilled]]: 0, [[elementSize]]: 1, [[ctor]]: %Uint8Array%, [[readerType]]: "default"}.
    4. Append pullIntoDescriptor as the last element of this.[[pendingPullIntos]].
  6. Let promise be ! ReadableStreamAddReadRequest(stream).
  7. Perform ! ReadableByteStreamControllerCallPullIfNeeded(this).
  8. Return promise.

3.11. Class ReadableStreamBYOBRequest

The ReadableStreamBYOBRequest class represents a pull into request in a ReadableByteStreamController.

3.11.1. Class Definition

This section is non-normative.

If one were to write the ReadableStreamBYOBRequest class in something close to the syntax of [ECMASCRIPT], it would look like

class ReadableStreamBYOBRequest {
  constructor(controller, view)

  get view()

  respond(bytesWritten)
  respondWithNewView(view)
}

3.11.2. Internal Slots

Instances of ReadableStreamBYOBRequest are created with the internal slots described in the following table:

Internal Slot Description (non-normative)
[[associatedReadableByteStreamController]] The parent ReadableByteStreamController instance
[[view]] A typed array representing the destination region to which the controller can write generated data

3.11.3. new ReadableStreamBYOBRequest(controller, view)

  1. If ! IsReadableByteStreamController(controller) is false, throw a TypeError exception.
  2. Set this.[[associatedReadableByteStreamController]] to controller.
  3. Set this.[[view]] to view.

3.11.4. Properties of the ReadableStreamBYOBRequest Prototype

3.11.4.1. get view
  1. If ! IsReadableStreamBYOBRequest(this) is false, throw a TypeError exception.
  2. Return this.[[view]].
3.11.4.2. respond(bytesWritten)
  1. If ! IsReadableStreamBYOBRequest(this) is false, throw a TypeError exception.
  2. If this.[[associatedReadableByteStreamController]] is undefined, throw a TypeError exception.
  3. If ! IsDetachedBuffer(this.[[view]].[[ViewedArrayBuffer]]) is true, throw a TypeError exception.
  4. Return ? ReadableByteStreamControllerRespond(this.[[associatedReadableByteStreamController]], bytesWritten).
3.11.4.3. respondWithNewView(view)
  1. If ! IsReadableStreamBYOBRequest(this) is false, throw a TypeError exception.
  2. If this.[[associatedReadableByteStreamController]] is undefined, throw a TypeError exception.
  3. If Type(view) is not Object, throw a TypeError exception.
  4. If view does not have a [[ViewedArrayBuffer]] internal slot, throw a TypeError exception.
  5. If ! IsDetachedBuffer(view.[[ViewedArrayBuffer]]) is true, throw a TypeError exception.
  6. Return ? ReadableByteStreamControllerRespondWithNewView(this.[[associatedReadableByteStreamController]], view).

3.12. Readable Stream BYOB Controller Abstract Operations

3.12.1. IsReadableStreamBYOBRequest ( x ) nothrow

  1. If Type(x) is not Object, return false.
  2. If x does not have an [[associatedReadableByteStreamController]] internal slot, return false.
  3. Return true.

3.12.2. IsReadableByteStreamController ( x ) nothrow

  1. If Type(x) is not Object, return false.
  2. If x does not have an [[controlledReadableByteStream]] internal slot, return false.
  3. Return true.

3.12.3. ReadableByteStreamControllerCallPullIfNeeded ( controller ) nothrow

  1. Let shouldPull be ! ReadableByteStreamControllerShouldCallPull(controller).
  2. If shouldPull is false, return.
  3. If controller.[[pulling]] is true,
    1. Set controller.[[pullAgain]] to true.
    2. Return.
  4. Assert: controller.[[pullAgain]] is false.
  5. Set controller.[[pulling]] to true.
  6. Let pullPromise be the result of performing controller.[[pullAlgorithm]].
  7. Upon fulfillment of pullPromise,
    1. Set controller.[[pulling]] to false.
    2. If controller.[[pullAgain]] is true,
      1. Set controller.[[pullAgain]] to false.
      2. Perform ! ReadableByteStreamControllerCallPullIfNeeded(controller).
  8. Upon rejection of pullPromise with reason e,
    1. If controller.[[controlledReadableByteStream]].[[state]] is "readable", perform ! ReadableByteStreamControllerError(controller, e).

3.12.4. ReadableByteStreamControllerClearPendingPullIntos ( controller ) nothrow

  1. Perform ! ReadableByteStreamControllerInvalidateBYOBRequest(controller).
  2. Set controller.[[pendingPullIntos]] to a new empty List.

3.12.5. ReadableByteStreamControllerClose ( controller ) throws

  1. Let stream be controller.[[controlledReadableByteStream]].
  2. Assert: controller.[[closeRequested]] is false.
  3. Assert: stream.[[state]] is "readable".
  4. If controller.[[queueTotalSize]] > 0,
    1. Set controller.[[closeRequested]] to true.
    2. Return.
  5. If controller.[[pendingPullIntos]] is not empty,
    1. Let firstPendingPullInto be the first element of controller.[[pendingPullIntos]].
    2. If firstPendingPullInto.[[bytesFilled]] > 0,
      1. Let e be a new TypeError exception.
      2. Perform ! ReadableByteStreamControllerError(controller, e).
      3. Throw e.
  6. Perform ! ReadableStreamClose(stream).

3.12.6. ReadableByteStreamControllerCommitPullIntoDescriptor ( stream, pullIntoDescriptor ) nothrow

  1. Assert: stream.[[state]] is not "errored".
  2. Let done be false.
  3. If stream.[[state]] is "closed",
    1. Assert: pullIntoDescriptor.[[bytesFilled]] is 0.
    2. Set done to true.
  4. Let filledView be ! ReadableByteStreamControllerConvertPullIntoDescriptor(pullIntoDescriptor).
  5. If pullIntoDescriptor.[[readerType]] is "default",
    1. Perform ! ReadableStreamFulfillReadRequest(stream, filledView, done).
  6. Otherwise,
    1. Assert: pullIntoDescriptor.[[readerType]] is "byob".
    2. Perform ! ReadableStreamFulfillReadIntoRequest(stream, filledView, done).

3.12.7. ReadableByteStreamControllerConvertPullIntoDescriptor ( pullIntoDescriptor ) nothrow

  1. Let bytesFilled be pullIntoDescriptor.[[bytesFilled]].
  2. Let elementSize be pullIntoDescriptor.[[elementSize]].
  3. Assert: bytesFilledpullIntoDescriptor.[[byteLength]].
  4. Assert: bytesFilled mod elementSize is 0.
  5. Return ! Construct(pullIntoDescriptor.[[ctor]], « pullIntoDescriptor.[[buffer]], pullIntoDescriptor.[[byteOffset]], bytesFilled ÷ elementSize »).

3.12.8. ReadableByteStreamControllerEnqueue ( controller, chunk ) nothrow

  1. Let stream be controller.[[controlledReadableByteStream]].
  2. Assert: controller.[[closeRequested]] is false.
  3. Assert: stream.[[state]] is "readable".
  4. Let buffer be chunk.[[ViewedArrayBuffer]].
  5. Let byteOffset be chunk.[[ByteOffset]].
  6. Let byteLength be chunk.[[ByteLength]].
  7. Let transferredBuffer be ! TransferArrayBuffer(buffer).
  8. If ! ReadableStreamHasDefaultReader(stream) is true
    1. If ! ReadableStreamGetNumReadRequests(stream) is 0,
      1. Perform ! ReadableByteStreamControllerEnqueueChunkToQueue(controller, transferredBuffer, byteOffset, byteLength).
    2. Otherwise,
      1. Assert: controller.[[queue]] is empty.
      2. Let transferredView be ! Construct(%Uint8Array%, « transferredBuffer, byteOffset, byteLength »).
      3. Perform ! ReadableStreamFulfillReadRequest(stream, transferredView, false).
  9. Otherwise, if ! ReadableStreamHasBYOBReader(stream) is true,
    1. Perform ! ReadableByteStreamControllerEnqueueChunkToQueue(controller, transferredBuffer, byteOffset, byteLength).
    2. Perform ! ReadableByteStreamControllerProcessPullIntoDescriptorsUsingQueue(controller).
  10. Otherwise,
    1. Assert: ! IsReadableStreamLocked(stream) is false.
    2. Perform ! ReadableByteStreamControllerEnqueueChunkToQueue(controller, transferredBuffer, byteOffset, byteLength).

3.12.9. ReadableByteStreamControllerEnqueueChunkToQueue ( controller, buffer, byteOffset, byteLength ) nothrow

  1. Append Record {[[buffer]]: buffer, [[byteOffset]]: byteOffset, [[byteLength]]: byteLength} as the last element of controller.[[queue]].
  2. Add byteLength to controller.[[queueTotalSize]].

3.12.10. ReadableByteStreamControllerError ( controller, e ) nothrow

  1. Let stream be controller.[[controlledReadableByteStream]].
  2. Assert: stream.[[state]] is "readable".
  3. Perform ! ReadableByteStreamControllerClearPendingPullIntos(controller).
  4. Perform ! ResetQueue(controller).
  5. Perform ! ReadableStreamError(stream, e).

3.12.11. ReadableByteStreamControllerFillHeadPullIntoDescriptor ( controller, size, pullIntoDescriptor ) nothrow

  1. Assert: either controller.[[pendingPullIntos]] is empty, or the first element of controller.[[pendingPullIntos]] is pullIntoDescriptor.
  2. Perform ! ReadableByteStreamControllerInvalidateBYOBRequest(controller).
  3. Set pullIntoDescriptor.[[bytesFilled]] to pullIntoDescriptor.[[bytesFilled]] + size.

3.12.12. ReadableByteStreamControllerFillPullIntoDescriptorFromQueue ( controller, pullIntoDescriptor ) nothrow

  1. Let elementSize be pullIntoDescriptor.[[elementSize]].
  2. Let currentAlignedBytes be pullIntoDescriptor.[[bytesFilled]] − (pullIntoDescriptor.[[bytesFilled]] mod elementSize).
  3. Let maxBytesToCopy be min(controller.[[queueTotalSize]], pullIntoDescriptor.[[byteLength]] − pullIntoDescriptor.[[bytesFilled]]).
  4. Let maxBytesFilled be pullIntoDescriptor.[[bytesFilled]] + maxBytesToCopy.
  5. Let maxAlignedBytes be maxBytesFilled − (maxBytesFilled mod elementSize).
  6. Let totalBytesToCopyRemaining be maxBytesToCopy.
  7. Let ready be false.
  8. If maxAlignedBytes > currentAlignedBytes,
    1. Set totalBytesToCopyRemaining to maxAlignedBytespullIntoDescriptor.[[bytesFilled]].
    2. Set ready to true.
  9. Let queue be controller.[[queue]].
  10. Repeat the following steps while totalBytesToCopyRemaining > 0,
    1. Let headOfQueue be the first element of queue.
    2. Let bytesToCopy be min(totalBytesToCopyRemaining, headOfQueue.[[byteLength]]).
    3. Let destStart be pullIntoDescriptor.[[byteOffset]] + pullIntoDescriptor.[[bytesFilled]].
    4. Perform ! CopyDataBlockBytes(pullIntoDescriptor.[[buffer]].[[ArrayBufferData]], destStart, headOfQueue.[[buffer]].[[ArrayBufferData]], headOfQueue.[[byteOffset]], bytesToCopy).
    5. If headOfQueue.[[byteLength]] is bytesToCopy,
      1. Remove the first element of queue, shifting all other elements downward (so that the second becomes the first, and so on).
    6. Otherwise,
      1. Set headOfQueue.[[byteOffset]] to headOfQueue.[[byteOffset]] + bytesToCopy.
      2. Set headOfQueue.[[byteLength]] to headOfQueue.[[byteLength]] − bytesToCopy.
    7. Set controller.[[queueTotalSize]] to controller.[[queueTotalSize]] − bytesToCopy.
    8. Perform ! ReadableByteStreamControllerFillHeadPullIntoDescriptor(controller, bytesToCopy, pullIntoDescriptor).
    9. Set totalBytesToCopyRemaining to totalBytesToCopyRemainingbytesToCopy.
  11. If ready is false,
    1. Assert: controller.[[queueTotalSize]] is 0.
    2. Assert: pullIntoDescriptor.[[bytesFilled]] > 0.
    3. Assert: pullIntoDescriptor.[[bytesFilled]] < pullIntoDescriptor.[[elementSize]].
  12. Return ready.

3.12.13. ReadableByteStreamControllerGetDesiredSize ( controller ) nothrow

  1. Let stream be controller.[[controlledReadableByteStream]].
  2. Let state be stream.[[state]].
  3. If state is "errored", return null.
  4. If state is "closed", return 0.
  5. Return controller.[[strategyHWM]] − controller.[[queueTotalSize]].

3.12.14. ReadableByteStreamControllerHandleQueueDrain ( controller ) nothrow

  1. Assert: controller.[[controlledReadableByteStream]].[[state]] is "readable".
  2. If controller.[[queueTotalSize]] is 0 and controller.[[closeRequested]] is true,
    1. Perform ! ReadableStreamClose(controller.[[controlledReadableByteStream]]).
  3. Otherwise,
    1. Perform ! ReadableByteStreamControllerCallPullIfNeeded(controller).

3.12.15. ReadableByteStreamControllerInvalidateBYOBRequest ( controller ) nothrow

  1. If controller.[[byobRequest]] is undefined, return.
  2. Set controller.[[byobRequest]].[[associatedReadableByteStreamController]] to undefined.
  3. Set controller.[[byobRequest]].[[view]] to undefined.
  4. Set controller.[[byobRequest]] to undefined.

3.12.16. ReadableByteStreamControllerProcessPullIntoDescriptorsUsingQueue ( controller ) nothrow

  1. Assert: controller.[[closeRequested]] is false.
  2. Repeat the following steps while controller.[[pendingPullIntos]] is not empty,
    1. If controller.[[queueTotalSize]] is 0, return.
    2. Let pullIntoDescriptor be the first element of controller.[[pendingPullIntos]].
    3. If ! ReadableByteStreamControllerFillPullIntoDescriptorFromQueue(controller, pullIntoDescriptor) is true,
      1. Perform ! ReadableByteStreamControllerShiftPendingPullInto(controller).
      2. Perform ! ReadableByteStreamControllerCommitPullIntoDescriptor(controller.[[controlledReadableByteStream]], pullIntoDescriptor).

3.12.17. ReadableByteStreamControllerPullInto ( controller, view ) nothrow

  1. Let stream be controller.[[controlledReadableByteStream]].
  2. Let elementSize be 1.
  3. Let ctor be %DataView%.
  4. If view has a [[TypedArrayName]] internal slot (i.e., it is not a DataView),
    1. Set elementSize to the element size specified in the typed array constructors table for view.[[TypedArrayName]].
    2. Set ctor to the constructor specified in the typed array constructors table for view.[[TypedArrayName]].
  5. Let buffer be ! TransferArrayBuffer(view.[[ViewedArrayBuffer]]).
  6. Let pullIntoDescriptor be Record {[[buffer]]: buffer, [[byteOffset]]: view.[[ByteOffset]], [[byteLength]]: view.[[ByteLength]], [[bytesFilled]]: 0, [[elementSize]]: elementSize, [[ctor]]: ctor, [[readerType]]: "byob"}.
  7. If controller.[[pendingPullIntos]] is not empty,
    1. Append pullIntoDescriptor as the last element of controller.[[pendingPullIntos]].
    2. Return ! ReadableStreamAddReadIntoRequest(stream).
  8. If stream.[[state]] is "closed",
    1. Let emptyView be ! Construct(ctor, « pullIntoDescriptor.[[buffer]], pullIntoDescriptor.[[byteOffset]], 0 »).
    2. Return a promise resolved with ! CreateIterResultObject(emptyView, true).
  9. If controller.[[queueTotalSize]] > 0,
    1. If ! ReadableByteStreamControllerFillPullIntoDescriptorFromQueue(controller, pullIntoDescriptor) is true,
      1. Let filledView be ! ReadableByteStreamControllerConvertPullIntoDescriptor(pullIntoDescriptor).
      2. Perform ! ReadableByteStreamControllerHandleQueueDrain(controller).
      3. Return a promise resolved with ! CreateIterResultObject(filledView, false).
    2. If controller.[[closeRequested]] is true,
      1. Let e be a TypeError exception.
      2. Perform ! ReadableByteStreamControllerError(controller, e).
      3. Return a promise rejected with e.
  10. Append pullIntoDescriptor as the last element of controller.[[pendingPullIntos]].
  11. Let promise be ! ReadableStreamAddReadIntoRequest(stream).
  12. Perform ! ReadableByteStreamControllerCallPullIfNeeded(controller).
  13. Return promise.

3.12.18. ReadableByteStreamControllerRespond ( controller, bytesWritten ) throws

  1. Let bytesWritten be ? ToNumber(bytesWritten).
  2. If ! IsFiniteNonNegativeNumber(bytesWritten) is false,
    1. Throw a RangeError exception.
  3. Assert: controller.[[pendingPullIntos]] is not empty.
  4. Perform ? ReadableByteStreamControllerRespondInternal(controller, bytesWritten).

3.12.19. ReadableByteStreamControllerRespondInClosedState ( controller, firstDescriptor ) nothrow

  1. Set firstDescriptor.[[buffer]] to ! TransferArrayBuffer(firstDescriptor.[[buffer]]).
  2. Assert: firstDescriptor.[[bytesFilled]] is 0.
  3. Let stream be controller.[[controlledReadableByteStream]].
  4. If ReadableStreamHasBYOBReader(stream) is true,
    1. Repeat the following steps while ! ReadableStreamGetNumReadIntoRequests(stream) > 0,
      1. Let pullIntoDescriptor be ! ReadableByteStreamControllerShiftPendingPullInto(controller).
      2. Perform ! ReadableByteStreamControllerCommitPullIntoDescriptor(stream, pullIntoDescriptor).

3.12.20. ReadableByteStreamControllerRespondInReadableState ( controller, bytesWritten, pullIntoDescriptor ) throws

  1. If pullIntoDescriptor.[[bytesFilled]] + bytesWritten > pullIntoDescriptor.[[byteLength]], throw a RangeError exception.
  2. Perform ! ReadableByteStreamControllerFillHeadPullIntoDescriptor(controller, bytesWritten, pullIntoDescriptor).
  3. If pullIntoDescriptor.[[bytesFilled]] < pullIntoDescriptor.[[elementSize]], return.
  4. Perform ! ReadableByteStreamControllerShiftPendingPullInto(controller).
  5. Let remainderSize be pullIntoDescriptor.[[bytesFilled]] mod pullIntoDescriptor.[[elementSize]].
  6. If remainderSize > 0,
    1. Let end be pullIntoDescriptor.[[byteOffset]] + pullIntoDescriptor.[[bytesFilled]].
    2. Let remainder be ? CloneArrayBuffer(pullIntoDescriptor.[[buffer]], endremainderSize, remainderSize, %ArrayBuffer%).
    3. Perform ! ReadableByteStreamControllerEnqueueChunkToQueue(controller, remainder, 0, remainder.[[ByteLength]]).
  7. Set pullIntoDescriptor.[[buffer]] to ! TransferArrayBuffer(pullIntoDescriptor.[[buffer]]).
  8. Set pullIntoDescriptor.[[bytesFilled]] to pullIntoDescriptor.[[bytesFilled]] − remainderSize.
  9. Perform ! ReadableByteStreamControllerCommitPullIntoDescriptor(controller.[[controlledReadableByteStream]], pullIntoDescriptor).
  10. Perform ! ReadableByteStreamControllerProcessPullIntoDescriptorsUsingQueue(controller).

3.12.21. ReadableByteStreamControllerRespondInternal ( controller, bytesWritten ) throws

  1. Let firstDescriptor be the first element of controller.[[pendingPullIntos]].
  2. Let stream be controller.[[controlledReadableByteStream]].
  3. If stream.[[state]] is "closed",
    1. If bytesWritten is not 0, throw a TypeError exception.
    2. Perform ! ReadableByteStreamControllerRespondInClosedState(controller, firstDescriptor).
  4. Otherwise,
    1. Assert: stream.[[state]] is "readable".
    2. Perform ? ReadableByteStreamControllerRespondInReadableState(controller, bytesWritten, firstDescriptor).

3.12.22. ReadableByteStreamControllerRespondWithNewView ( controller, view ) throws

  1. Assert: controller.[[pendingPullIntos]] is not empty.
  2. Let firstDescriptor be the first element of controller.[[pendingPullIntos]].
  3. If firstDescriptor.[[byteOffset]] + firstDescriptor.[[bytesFilled]] is not view.[[ByteOffset]], throw a RangeError exception.
  4. If firstDescriptor.[[byteLength]] is not view.[[ByteLength]], throw a RangeError exception.
  5. Set firstDescriptor.[[buffer]] to view.[[ViewedArrayBuffer]].
  6. Perform ? ReadableByteStreamControllerRespondInternal(controller, view.[[ByteLength]]).

3.12.23. ReadableByteStreamControllerShiftPendingPullInto ( controller ) nothrow

  1. Let descriptor be the first element of controller.[[pendingPullIntos]].
  2. Remove descriptor from controller.[[pendingPullIntos]], shifting all other elements downward (so that the second becomes the first, and so on).
  3. Perform ! ReadableByteStreamControllerInvalidateBYOBRequest(controller).
  4. Return descriptor.

3.12.24. ReadableByteStreamControllerShouldCallPull ( controller ) nothrow

  1. Let stream be controller.[[controlledReadableByteStream]].
  2. If stream.[[state]] is not "readable", return false.
  3. If controller.[[closeRequested]] is true, return false.
  4. If controller.[[started]] is false, return false.
  5. If ! ReadableStreamHasDefaultReader(stream) is true and ! ReadableStreamGetNumReadRequests(stream) > 0, return true.
  6. If ! ReadableStreamHasBYOBReader(stream) is true and ! ReadableStreamGetNumReadIntoRequests(stream) > 0, return true.
  7. If ! ReadableByteStreamControllerGetDesiredSize(controller) > 0, return true.
  8. Return false.

3.12.25. SetUpReadableByteStreamController ( stream, controller, startAlgorithm, pullAlgorithm, cancelAlgorithm, highWaterMark, autoAllocateChunkSize ) throws

  1. Assert: stream.[[readableStreamController]] is undefined.
  2. If autoAllocateChunkSize is not undefined,
    1. Assert: ! IsInteger(autoAllocateChunkSize) is true.
    2. Assert: autoAllocateChunkSize is positive.
  3. Set controller.[[controlledReadableByteStream]] to stream.
  4. Set controller.[[pullAgain]] and controller.[[pulling]] to false.
  5. Perform ! ReadableByteStreamControllerClearPendingPullIntos(controller).
  6. Perform ! ResetQueue(controller).
  7. Set controller.[[closeRequested]] and controller.[[started]] to false.
  8. Set controller.[[strategyHWM]] to ? ValidateAndNormalizeHighWaterMark(highWaterMark).
  9. Set controller.[[pullAlgorithm]] to pullAlgorithm.
  10. Set controller.[[cancelAlgorithm]] to cancelAlgorithm.
  11. Set controller.[[autoAllocateChunkSize]] to autoAllocateChunkSize.
  12. Set controller.[[pendingPullIntos]] to a new empty List.
  13. Set stream.[[readableStreamController]] to controller.
  14. Let startResult be the result of performing startAlgorithm.
  15. Let startPromise be a promise resolved with startResult.
  16. Upon fulfillment of startPromise,
    1. Set controller.[[started]] to true.
    2. Assert: controller.[[pulling]] is false.
    3. Assert: controller.[[pullAgain]] is false.
    4. Perform ! ReadableByteStreamControllerCallPullIfNeeded(controller).
  17. Upon rejection of startPromise with reason r,
    1. If stream.[[state]] is "readable", perform ! ReadableByteStreamControllerError(controller, r).

3.12.26. SetUpReadableByteStreamControllerFromUnderlyingSource ( stream, underlyingByteSource, highWaterMark ) throws

  1. Assert: underlyingByteSource is not undefined.
  2. Let controller be ObjectCreate(the original value of ReadableByteStreamController's prototype property).
  3. Let startAlgorithm be the following steps:
    1. Return ? InvokeOrNoop(underlyingByteSource, "start", « controller »).
  4. Let pullAlgorithm be ? CreateAlgorithmFromUnderlyingMethod(underlyingByteSource, "pull", 0, « controller »).
  5. Let cancelAlgorithm be ? CreateAlgorithmFromUnderlyingMethod(underlyingByteSource, "cancel", 1, « »).
  6. Let autoAllocateChunkSize be ? GetV(underlyingByteSource, "autoAllocateChunkSize").
  7. If autoAllocateChunkSize is not undefined,
    1. If ! IsInteger(autoAllocateChunkSize) is false, or if autoAllocateChunkSize0, throw a RangeError exception.
  8. Perform ? SetUpReadableByteStreamController(stream, controller, startAlgorithm, pullAlgorithm, cancelAlgorithm, highWaterMark, autoAllocateChunkSize).

4. Writable Streams

4.1. Using Writable Streams

The usual way to write to a writable stream is to simply pipe a readable stream to it. This ensures that backpressure is respected, so that if the writable stream’s underlying sink is not able to accept data as fast as the readable stream can produce it, the readable stream is informed of this and has a chance to slow down its data production.
readableStream.pipeTo(writableStream)
  .then(() => console.log("All data successfully written!"))
  .catch(e => console.error("Something went wrong!", e));
You can also write directly to writable streams by acquiring a writer and using its write() and close() methods. Since writable streams queue any incoming writes, and take care internally to forward them to the underlying sink in sequence, you can indiscriminately write to a writable stream without much ceremony:
function writeArrayToStream(array, writableStream) {
  const writer = writableStream.getWriter();
  array.forEach(chunk => writer.write(chunk));

  return writer.close();
}

writeArrayToStream([1, 2, 3, 4, 5], writableStream)
  .then(() => console.log("All done!"))
  .catch(e => console.error("Error with the stream: " + e));
In the previous example we only paid attention to the success or failure of the entire stream, by looking at the promise returned by the writer’s close() method. That promise will reject if anything goes wrong with the stream—initializing it, writing to it, or closing it. And it will fulfill once the stream is successfully closed. Often this is all you care about.

However, if you care about the success of writing a specific chunk, you can use the promise returned by the writer’s write() method:

writer.write("i am a chunk of data")
  .then(() => console.log("chunk successfully written!"))
  .catch(e => console.error(e));

What "success" means is up to a given stream instance (or more precisely, its underlying sink) to decide. For example, for a file stream it could simply mean that the OS has accepted the write, and not necessarily that the chunk has been flushed to disk.

The desiredSize and ready properties of writable stream writers allow producers to more precisely respond to flow control signals from the stream, to keep memory usage below the stream’s specified high water mark. The following example writes an infinite sequence of random bytes to a stream, using desiredSize to determine how many bytes to generate at a given time, and using ready to wait for the backpressure to subside.
async function writeRandomBytesForever(writableStream) {
  const writer = writableStream.getWriter();

  while (true) {
    await writer.ready;

    const bytes = new Uint8Array(writer.desiredSize);
    window.crypto.getRandomValues(bytes);

    await writer.write(bytes);
  }
}

writeRandomBytesForever(myWritableStream).catch(e => console.error("Something broke", e));

4.2. Class WritableStream

4.2.1. Class Definition

This section is non-normative.

If one were to write the WritableStream class in something close to the syntax of [ECMASCRIPT], it would look like

class WritableStream {
  constructor(underlyingSink = {}, { size, highWaterMark = 1 } = {})

  get locked()

  abort(reason)
  getWriter()
}

4.2.2. Internal Slots

Instances of WritableStream are created with the internal slots described in the following table:

Internal Slot Description (non-normative)
[[backpressure]] The backpressure signal set by the controller
[[closeRequest]] The promise returned from the writer close() method
[[inFlightWriteRequest]] A slot set to the promise for the current in-flight write operation while the underlying sink’s write algorithm is executing and has not yet fulfilled, used to prevent reentrant calls
[[inFlightCloseRequest]] A slot set to the promise for the current in-flight close operation while the underlying sink’s close algorithm is executing and has not yet fulfilled, used to prevent the abort() method from interrupting close
[[pendingAbortRequest]] A Record containing the promise returned from abort() and the reason passed to abort()
[[state]] A string containing the stream’s current state, used internally; one of "writable", "closed", "erroring", or "errored"
[[storedError]] A value indicating how the stream failed, to be given as a failure reason or exception when trying to operate on the stream while in the "errored" state
[[writableStreamController]] A WritableStreamDefaultController created with the ability to control the state and queue of this stream; also used for the IsWritableStream brand check
[[writer]] A WritableStreamDefaultWriter instance, if the stream is locked to a writer, or undefined if it is not
[[writeRequests]] A List of promises representing the stream’s internal queue of write requests not yet processed by the underlying sink

The [[inFlightCloseRequest]] slot and [[closeRequest]] slot are mutually exclusive. Similarly, no element will be removed from [[writeRequests]] while [[inFlightWriteRequest]] is not undefined. Implementations can optimize storage for these slots based on these invariants.

4.2.3. new WritableStream(underlyingSink = {}, { size, highWaterMark = 1 } = {})

The underlyingSink object passed to the constructor can implement any of the following methods to govern how the constructed stream instance behaves:

The controller object passed to start, write and close is an instance of WritableStreamDefaultController, and has the ability to error the stream.

The constructor also accepts a second argument containing the queuing strategy object with two properties: a non-negative number highWaterMark, and a function size(chunk). The supplied strategy could be an instance of the built-in CountQueuingStrategy or ByteLengthQueuingStrategy classes, or it could be custom. If no strategy is supplied, the default behavior will be the same as a CountQueuingStrategy with a high water mark of 1.

  1. Perform ! InitializeWritableStream(this).
  2. Let type be ? GetV(underlyingSink, "type").
  3. If type is not undefined, throw a RangeError exception.

    This is to allow us to add new potential types in the future, without backward-compatibility concerns.

  4. Let sizeAlgorithm be ? MakeSizeAlgorithmFromSizeFunction(size).
  5. Set highWaterMark to ? ValidateAndNormalizeHighWaterMark(highWaterMark).
  6. Perform ? SetUpWritableStreamDefaultControllerFromUnderlyingSink(this, underlyingSink, highWaterMark, sizeAlgorithm).

4.2.4. Properties of the WritableStream Prototype

4.2.4.1. get locked
The locked getter returns whether or not the writable stream is locked to a writer.
  1. If ! IsWritableStream(this) is false, throw a TypeError exception.
  2. Return ! IsWritableStreamLocked(this).
4.2.4.2. abort(reason)
The abort method aborts the stream, signaling that the producer can no longer successfully write to the stream and it is to be immediately moved to an errored state, with any queued-up writes discarded. This will also execute any abort mechanism of the underlying sink.
  1. If ! IsWritableStream(this) is false, return a promise rejected with a TypeError exception.
  2. If ! IsWritableStreamLocked(this) is true, return a promise rejected with a TypeError exception.
  3. Return ! WritableStreamAbort(this, reason).
4.2.4.3. getWriter()
The getWriter method creates a writer (an instance of WritableStreamDefaultWriter) and locks the stream to the new writer. While the stream is locked, no other writer can be acquired until this one is released.

This functionality is especially useful for creating abstractions that desire the ability to write to a stream without interruption or interleaving. By getting a writer for the stream, you can ensure nobody else can write at the same time, which would cause the resulting written data to be unpredictable and probably useless.

  1. If ! IsWritableStream(this) is false, throw a TypeError exception.
  2. Return ? AcquireWritableStreamDefaultWriter(this).

4.3. General Writable Stream Abstract Operations

The following abstract operations, unlike most in this specification, are meant to be generally useful by other specifications, instead of just being part of the implementation of this spec’s classes.

4.3.1. AcquireWritableStreamDefaultWriter ( stream ) throws

  1. Return ? Construct(WritableStreamDefaultWriter, « stream »).

4.3.2. CreateWritableStream ( startAlgorithm, writeAlgorithm, closeAlgorithm, abortAlgorithm [, highWaterMark [, sizeAlgorithm ] ] ) throws

This abstract operation is meant to be called from other specifications that wish to create WritableStream instances. The writeAlgorithm, closeAlgorithm and abortAlgorithm algorithms must return promises; if supplied, sizeAlgorithm must be an algorithm accepting chunk objects and returning a number; and if supplied, highWaterMark must be a nonnegative, non-NaN number.

CreateWritableStream throws an exception if and only if the supplied startAlgorithm throws.

  1. If highWaterMark was not passed, set it to 1.
  2. If sizeAlgorithm was not passed, set it to an algorithm that returns 1.
  3. Assert: ! IsNonNegativeNumber(highWaterMark) is true.
  4. Let stream be ObjectCreate(the original value of WritableStream's prototype property).
  5. Perform ! InitializeWritableStream(stream).
  6. Let controller be ObjectCreate(the original value of WritableStreamDefaultController's prototype property).
  7. Perform ? SetUpWritableStreamDefaultController(stream, controller, startAlgorithm, writeAlgorithm, closeAlgorithm, abortAlgorithm, highWaterMark, sizeAlgorithm).
  8. Return stream.

4.3.3. InitializeWritableStream ( stream ) nothrow

  1. Set stream.[[state]] to "writable".
  2. Set stream.[[storedError]], stream.[[writer]], stream.[[writableStreamController]], stream.[[inFlightWriteRequest]], stream.[[closeRequest]], stream.[[inFlightCloseRequest]] and stream.[[pendingAbortRequest]] to undefined.
  3. Set stream.[[writeRequests]] to a new empty List.
  4. Set stream.[[backpressure]] to false.

4.3.4. IsWritableStream ( x ) nothrow

  1. If Type(x) is not Object, return false.
  2. If x does not have a [[writableStreamController]] internal slot, return false.
  3. Return true.

4.3.5. IsWritableStreamLocked ( stream ) nothrow

This abstract operation is meant to be called from other specifications that may wish to query whether or not a writable stream is locked to a writer.

  1. Assert: ! IsWritableStream(stream) is true.
  2. If stream.[[writer]] is undefined, return false.
  3. Return true.

4.3.6. WritableStreamAbort ( stream, reason ) nothrow

  1. Let state be stream.[[state]].
  2. If state is "closed", return a promise resolved with undefined.
  3. If state is "errored", return a promise rejected with stream.[[storedError]].
  4. Let error be a new TypeError indicating that the stream has been requested to abort.
  5. If stream.[[pendingAbortRequest]] is not undefined, return a promise rejected with error.
  6. Assert: state is "writable" or "erroring".
  7. Let wasAlreadyErroring be false.
  8. If state is "erroring",
    1. Set wasAlreadyErroring to true.
    2. Set reason to undefined.
  9. Let promise be a new promise.
  10. Set stream.[[pendingAbortRequest]] to Record {[[promise]]: promise, [[reason]]: reason, [[wasAlreadyErroring]]: wasAlreadyErroring}.
  11. If wasAlreadyErroring is false, perform ! WritableStreamStartErroring(stream, error).
  12. Return promise.

4.4. Writable Stream Abstract Operations Used by Controllers

To allow future flexibility to add different writable stream behaviors (similar to the distinction between default readable streams and readable byte streams), much of the internal state of a writable stream is encapsulated by the WritableStreamDefaultController class.

The abstract operations in this section are interfaces that are used by the controller implementation to affect its associated WritableStream object, translating the controller’s internal state changes into developer-facing results visible through the WritableStream's public API.

4.4.1. WritableStreamAddWriteRequest ( stream ) nothrow

  1. Assert: ! IsWritableStreamLocked(stream) is true.
  2. Assert: stream.[[state]] is "writable".
  3. Let promise be a new promise.
  4. Append promise as the last element of stream.[[writeRequests]].
  5. Return promise.

4.4.2. WritableStreamDealWithRejection ( stream, error ) nothrow

  1. Let state be stream.[[state]].
  2. If state is "writable",
    1. Perform ! WritableStreamStartErroring(stream, error).
    2. Return.
  3. Assert: state is "erroring".
  4. Perform ! WritableStreamFinishErroring(stream).

4.4.3. WritableStreamStartErroring ( stream, reason ) nothrow

  1. Assert: stream.[[storedError]] is undefined.
  2. Assert: stream.[[state]] is "writable".
  3. Let controller be stream.[[writableStreamController]].
  4. Assert: controller is not undefined.
  5. Set stream.[[state]] to "erroring".
  6. Set stream.[[storedError]] to reason.
  7. Let writer be stream.[[writer]].
  8. If writer is not undefined, perform ! WritableStreamDefaultWriterEnsureReadyPromiseRejected(writer, reason).
  9. If ! WritableStreamHasOperationMarkedInFlight(stream) is false and controller.[[started]] is true, perform ! WritableStreamFinishErroring(stream).

4.4.4. WritableStreamFinishErroring ( stream ) nothrow

  1. Assert: stream.[[state]] is "erroring".
  2. Assert: ! WritableStreamHasOperationMarkedInFlight(stream) is false.
  3. Set stream.[[state]] to "errored".
  4. Perform ! stream.[[writableStreamController]].[[ErrorSteps]]().
  5. Let storedError be stream.[[storedError]].
  6. Repeat for each writeRequest that is an element of stream.[[writeRequests]],
    1. Reject writeRequest with storedError.
  7. Set stream.[[writeRequests]] to an empty List.
  8. if stream.[[pendingAbortRequest]] is undefined,
    1. Perform ! WritableStreamRejectCloseAndClosedPromiseIfNeeded(stream).
    2. Return.
  9. Let abortRequest be stream.[[pendingAbortRequest]].
  10. Set stream.[[pendingAbortRequest]] to undefined.
  11. If abortRequest.[[wasAlreadyErroring]] is true,
    1. Reject abortRequest.[[promise]] with storedError.
    2. Perform ! WritableStreamRejectCloseAndClosedPromiseIfNeeded(stream).
    3. Return.
  12. Let promise be ! stream.[[writableStreamController]].[[AbortSteps]](abortRequest.[[reason]]).
  13. Upon fulfillment of promise,
    1. Resolve abortRequest.[[promise]] with undefined.
    2. Perform ! WritableStreamRejectCloseAndClosedPromiseIfNeeded(stream).
  14. Upon rejection of promise with reason reason,
    1. Reject abortRequest.[[promise]] with reason.
    2. Perform ! WritableStreamRejectCloseAndClosedPromiseIfNeeded(stream).

4.4.5. WritableStreamFinishInFlightWrite ( stream ) nothrow

  1. Assert: stream.[[inFlightWriteRequest]] is not undefined.
  2. Resolve stream.[[inFlightWriteRequest]] with undefined.
  3. Set stream.[[inFlightWriteRequest]] to undefined.

4.4.6. WritableStreamFinishInFlightWriteWithError ( stream, error ) nothrow

  1. Assert: stream.[[inFlightWriteRequest]] is not undefined.
  2. Reject stream.[[inFlightWriteRequest]] with error.
  3. Set stream.[[inFlightWriteRequest]] to undefined.
  4. Assert: stream.[[state]] is "writable" or "erroring".
  5. Perform ! WritableStreamDealWithRejection(stream, error).

4.4.7. WritableStreamFinishInFlightClose ( stream ) nothrow

  1. Assert: stream.[[inFlightCloseRequest]] is not undefined.
  2. Resolve stream.[[inFlightCloseRequest]] with undefined.
  3. Set stream.[[inFlightCloseRequest]] to undefined.
  4. Let state be stream.[[state]].
  5. Assert: stream.[[state]] is "writable" or "erroring".
  6. If state is "erroring",
    1. Set stream.[[storedError]] to undefined.
    2. If stream.[[pendingAbortRequest]] is not undefined,
      1. Resolve stream.[[pendingAbortRequest]].[[promise]] with undefined.
      2. Set stream.[[pendingAbortRequest]] to undefined.
  7. Set stream.[[state]] to "closed".
  8. Let writer be stream.[[writer]].
  9. If writer is not undefined, resolve writer.[[closedPromise]] with undefined.
  10. Assert: stream.[[pendingAbortRequest]] is undefined.
  11. Assert: stream.[[storedError]] is undefined.

4.4.8. WritableStreamFinishInFlightCloseWithError ( stream, error ) nothrow

  1. Assert: stream.[[inFlightCloseRequest]] is not undefined.
  2. Reject stream.[[inFlightCloseRequest]] with error.
  3. Set stream.[[inFlightCloseRequest]] to undefined.
  4. Assert: stream.[[state]] is "writable" or "erroring".
  5. If stream.[[pendingAbortRequest]] is not undefined,
    1. Reject stream.[[pendingAbortRequest]].[[promise]] with error.
    2. Set stream.[[pendingAbortRequest]] to undefined.
  6. Perform ! WritableStreamDealWithRejection(stream, error).

4.4.9. WritableStreamCloseQueuedOrInFlight ( stream ) nothrow

  1. If stream.[[closeRequest]] is undefined and stream.[[inFlightCloseRequest]] is undefined, return false.
  2. Return true.

4.4.10. WritableStreamHasOperationMarkedInFlight ( stream ) nothrow

  1. If stream.[[inFlightWriteRequest]] is undefined and controller.[[inFlightCloseRequest]] is undefined, return false.
  2. Return true.

4.4.11. WritableStreamMarkCloseRequestInFlight ( stream ) nothrow

  1. Assert: stream.[[inFlightCloseRequest]] is undefined.
  2. Assert: stream.[[closeRequest]] is not undefined.
  3. Set stream.[[inFlightCloseRequest]] to stream.[[closeRequest]].
  4. Set stream.[[closeRequest]] to undefined.

4.4.12. WritableStreamMarkFirstWriteRequestInFlight ( stream ) nothrow

  1. Assert: stream.[[inFlightWriteRequest]] is undefined.
  2. Assert: stream.[[writeRequests]] is not empty.
  3. Let writeRequest be the first element of stream.[[writeRequests]].
  4. Remove writeRequest from stream.[[writeRequests]], shifting all other elements downward (so that the second becomes the first, and so on).
  5. Set stream.[[inFlightWriteRequest]] to writeRequest.

4.4.13. WritableStreamRejectCloseAndClosedPromiseIfNeeded ( stream ) nothrow

  1. Assert: stream.[[state]] is "errored".
  2. If stream.[[closeRequest]] is not undefined,
    1. Assert: stream.[[inFlightCloseRequest]] is undefined.
    2. Reject stream.[[closeRequest]] with stream.[[storedError]].
    3. Set stream.[[closeRequest]] to undefined.
  3. Let writer be stream.[[writer]].
  4. If writer is not undefined,
    1. Reject writer.[[closedPromise]] with stream.[[storedError]].
    2. Set writer.[[closedPromise]].[[PromiseIsHandled]] to true.

4.4.14. WritableStreamUpdateBackpressure ( stream, backpressure ) nothrow

  1. Assert: stream.[[state]] is "writable".
  2. Assert: ! WritableStreamCloseQueuedOrInFlight(stream) is false.
  3. Let writer be stream.[[writer]].
  4. If writer is not undefined and backpressure is not stream.[[backpressure]],
    1. If backpressure is true, set writer.[[readyPromise]] to a new promise.
    2. Otherwise,
      1. Assert: backpressure is false.
      2. Resolve writer.[[readyPromise]] with undefined.
  5. Set stream.[[backpressure]] to backpressure.

4.5. Class WritableStreamDefaultWriter

The WritableStreamDefaultWriter class represents a writable stream writer designed to be vended by a WritableStream instance.

4.5.1. Class Definition

This section is non-normative.

If one were to write the WritableStreamDefaultWriter class in something close to the syntax of [ECMASCRIPT], it would look like

class WritableStreamDefaultWriter {
  constructor(stream)

  get closed()
  get desiredSize()
  get ready()

  abort(reason)
  close()
  releaseLock()
  write(chunk)
}

4.5.2. Internal Slots

Instances of WritableStreamDefaultWriter are created with the internal slots described in the following table:

Internal Slot Description (non-normative)
[[closedPromise]] A promise returned by the writer’s closed getter
[[ownerWritableStream]] A WritableStream instance that owns this writer
[[readyPromise]] A promise returned by the writer’s ready getter

4.5.3. new WritableStreamDefaultWriter(stream)

The WritableStreamDefaultWriter constructor is generally not meant to be used directly; instead, a stream’s getWriter() method ought to be be used.
  1. If ! IsWritableStream(stream) is false, throw a TypeError exception.
  2. If ! IsWritableStreamLocked(stream) is true, throw a TypeError exception.
  3. Set this.[[ownerWritableStream]] to stream.
  4. Set stream.[[writer]] to this.
  5. Let state be stream.[[state]].
  6. If state is "writable",
    1. If ! WritableStreamCloseQueuedOrInFlight(stream) is false and stream.[[backpressure]] is true, set this.[[readyPromise]] to a new promise.
    2. Otherwise, set this.[[readyPromise]] to a promise resolved with undefined.
    3. Set this.[[closedPromise]] to a new promise.
  7. Otherwise, if state is "erroring",
    1. Set this.[[readyPromise]] to a promise rejected with stream.[[storedError]].
    2. Set this.[[readyPromise]].[[PromiseIsHandled]] to true.
    3. Set this.[[closedPromise]] to a new promise.
  8. Otherwise, if state is "closed",
    1. Set this.[[readyPromise]] to a promise resolved with undefined.
    2. Set this.[[closedPromise]] to a promise resolved with undefined.
  9. Otherwise,
    1. Assert: state is "errored".
    2. Let storedError be stream.[[storedError]].
    3. Set this.[[readyPromise]] to a promise rejected with storedError.
    4. Set this.[[readyPromise]].[[PromiseIsHandled]] to true.
    5. Set this.[[closedPromise]] to a promise rejected with storedError.
    6. Set this.[[closedPromise]].[[PromiseIsHandled]] to true.

4.5.4. Properties of the WritableStreamDefaultWriter Prototype

4.5.4.1. get closed
The closed getter returns a promise that will be fulfilled when the stream becomes closed, or rejected if the stream ever errors or the writer’s lock is released before the stream finishes closing.
  1. If ! IsWritableStreamDefaultWriter(this) is false, return a promise rejected with a TypeError exception.
  2. Return this.[[closedPromise]].
4.5.4.2. get desiredSize
The desiredSize getter returns the desired size to fill the stream’s internal queue. It can be negative, if the queue is over-full. A producer can use this information to determine the right amount of data to write.

It will be null if the stream cannot be successfully written to (due to either being errored, or having an abort queued up). It will return zero if the stream is closed. The getter will throw an exception if invoked when the writer’s lock is released.

  1. If ! IsWritableStreamDefaultWriter(this) is false, throw a TypeError exception.
  2. If this.[[ownerWritableStream]] is undefined, throw a TypeError exception.
  3. Return ! WritableStreamDefaultWriterGetDesiredSize(this).
4.5.4.3. get ready
The ready getter returns a promise that will be fulfilled when the desired size to fill the stream’s internal queue transitions from nonpositive to positive, signaling that it is no longer applying backpressure. Once the desired size to fill the stream’s internal queue dips back to zero or below, the getter will return a new promise that stays pending until the next transition.

If the stream becomes errored or aborted, or the writer’s lock is released, the returned promise will become rejected.

  1. If ! IsWritableStreamDefaultWriter(this) is false, return a promise rejected with a TypeError exception.
  2. Return this.[[readyPromise]].
4.5.4.4. abort(reason)
If the writer is active, the abort method behaves the same as that for the associated stream. (Otherwise, it returns a rejected promise.)
  1. If ! IsWritableStreamDefaultWriter(this) is false, return a promise rejected with a TypeError exception.
  2. If this.[[ownerWritableStream]] is undefined, return a promise rejected with a TypeError exception.
  3. Return ! WritableStreamDefaultWriterAbort(this, reason).
4.5.4.5. close()
The close method will close the associated writable stream. The underlying sink will finish processing any previously-written chunks, before invoking its close behavior. During this time any further attempts to write will fail (without erroring the stream).

The method returns a promise that is fulfilled with undefined if all remaining chunks are successfully written and the stream successfully closes, or rejects if an error is encountered during this process.

  1. If ! IsWritableStreamDefaultWriter(this) is false, return a promise rejected with a TypeError exception.
  2. Let stream be this.[[ownerWritableStream]].
  3. If stream is undefined, return a promise rejected with a TypeError exception.
  4. If ! WritableStreamCloseQueuedOrInFlight(stream) is true, return a promise rejected with a TypeError exception.
  5. Return ! WritableStreamDefaultWriterClose(this).
4.5.4.6. releaseLock()
The releaseLock method releases the writer’s lock on the corresponding stream. After the lock is released, the writer is no longer active. If the associated stream is errored when the lock is released, the writer will appear errored in the same way from now on; otherwise, the writer will appear closed.

Note that the lock can still be released even if some ongoing writes have not yet finished (i.e. even if the promises returned from previous calls to write() have not yet settled). It’s not necessary to hold the lock on the writer for the duration of the write; the lock instead simply prevents other producers from writing in an interleaved manner.

  1. If ! IsWritableStreamDefaultWriter(this) is false, throw a TypeError exception.
  2. Let stream be this.[[ownerWritableStream]].
  3. If stream is undefined, return.
  4. Assert: stream.[[writer]] is not undefined.
  5. Perform ! WritableStreamDefaultWriterRelease(this).
4.5.4.7. write(chunk)
The write method writes the given chunk to the writable stream, by waiting until any previous writes have finished successfully, and then sending the chunk to the underlying sink. It will return a promise that fulfills with undefined upon a successful write, or rejects if the write fails or stream becomes errored before the writing process is initiated.

Note that what "success" means is up to the underlying sink; it might indicate simply that the chunk has been accepted, and not necessarily that it is safely saved to its ultimate destination.

  1. If ! IsWritableStreamDefaultWriter(this) is false, return a promise rejected with a TypeError exception.
  2. If this.[[ownerWritableStream]] is undefined, return a promise rejected with a TypeError exception.
  3. Return ! WritableStreamDefaultWriterWrite(this, chunk).

4.6. Writable Stream Writer Abstract Operations

4.6.1. IsWritableStreamDefaultWriter ( x ) nothrow

  1. If Type(x) is not Object, return false.
  2. If x does not have an [[ownerWritableStream]] internal slot, return false.
  3. Return true.

4.6.2. WritableStreamDefaultWriterAbort ( writer, reason ) nothrow

  1. Let stream be writer.[[ownerWritableStream]].
  2. Assert: stream is not undefined.
  3. Return ! WritableStreamAbort(stream, reason).

4.6.3. WritableStreamDefaultWriterClose ( writer ) nothrow

  1. Let stream be writer.[[ownerWritableStream]].
  2. Assert: stream is not undefined.
  3. Let state be stream.[[state]].
  4. If state is "closed" or "errored", return a promise rejected with a TypeError exception.
  5. Assert: state is "writable" or "erroring".
  6. Assert: ! WritableStreamCloseQueuedOrInFlight(stream) is false.
  7. Let promise be a new promise.
  8. Set stream.[[closeRequest]] to promise.
  9. If stream.[[backpressure]] is true and state is "writable", resolve writer.[[readyPromise]] with undefined.
  10. Perform ! WritableStreamDefaultControllerClose(stream.[[writableStreamController]]).
  11. Return promise.

4.6.4. WritableStreamDefaultWriterCloseWithErrorPropagation ( writer ) nothrow

This abstract operation helps implement the error propagation semantics of pipeTo().

  1. Let stream be writer.[[ownerWritableStream]].
  2. Assert: stream is not undefined.
  3. Let state be stream.[[state]].
  4. If ! WritableStreamCloseQueuedOrInFlight(stream) is true or state is "closed", return a promise resolved with undefined.
  5. If state is "errored", return a promise rejected with stream.[[storedError]].
  6. Assert: state is "writable" or "erroring".
  7. Return ! WritableStreamDefaultWriterClose(writer).

4.6.5. WritableStreamDefaultWriterEnsureClosedPromiseRejected( writer, error ) nothrow

  1. If writer.[[closedPromise]].[[PromiseState]] is "pending", reject writer.[[closedPromise]] with error.
  2. Otherwise, set writer.[[closedPromise]] to a promise rejected with error.
  3. Set writer.[[closedPromise]].[[PromiseIsHandled]] to true.

4.6.6. WritableStreamDefaultWriterEnsureReadyPromiseRejected( writer, error ) nothrow

  1. If writer.[[readyPromise]].[[PromiseState]] is "pending", reject writer.[[readyPromise]] with error.
  2. Otherwise, set writer.[[readyPromise]] to a promise rejected with error.
  3. Set writer.[[readyPromise]].[[PromiseIsHandled]] to true.

4.6.7. WritableStreamDefaultWriterGetDesiredSize ( writer ) nothrow

  1. Let stream be writer.[[ownerWritableStream]].
  2. Let state be stream.[[state]].
  3. If state is "errored" or "erroring", return null.
  4. If state is "closed", return 0.
  5. Return ! WritableStreamDefaultControllerGetDesiredSize(stream.[[writableStreamController]]).

4.6.8. WritableStreamDefaultWriterRelease ( writer ) nothrow

  1. Let stream be writer.[[ownerWritableStream]].
  2. Assert: stream is not undefined.
  3. Assert: stream.[[writer]] is writer.
  4. Let releasedError be a new TypeError.
  5. Perform ! WritableStreamDefaultWriterEnsureReadyPromiseRejected(writer, releasedError).
  6. Perform ! WritableStreamDefaultWriterEnsureClosedPromiseRejected(writer, releasedError).
  7. Set stream.[[writer]] to undefined.
  8. Set writer.[[ownerWritableStream]] to undefined.

4.6.9. WritableStreamDefaultWriterWrite ( writer, chunk ) nothrow

  1. Let stream be writer.[[ownerWritableStream]].
  2. Assert: stream is not undefined.
  3. Let controller be stream.[[writableStreamController]].
  4. Let chunkSize be ! WritableStreamDefaultControllerGetChunkSize(controller, chunk).
  5. If stream is not equal to writer.[[ownerWritableStream]], return a promise rejected with a TypeError exception.
  6. Let state be stream.[[state]].
  7. If state is "errored", return a promise rejected with stream.[[storedError]].
  8. If ! WritableStreamCloseQueuedOrInFlight(stream) is true or state is "closed", return a promise rejected with a TypeError exception indicating that the stream is closing or closed.
  9. If state is "erroring", return a promise rejected with stream.[[storedError]].
  10. Assert: state is "writable".
  11. Let promise be ! WritableStreamAddWriteRequest(stream).
  12. Perform ! WritableStreamDefaultControllerWrite(controller, chunk, chunkSize).
  13. Return promise.

4.7. Class WritableStreamDefaultController

The WritableStreamDefaultController class has methods that allow control of a WritableStream's state. When constructing a WritableStream, the underlying sink is given a corresponding WritableStreamDefaultController instance to manipulate.

4.7.1. Class Definition

This section is non-normative.

If one were to write the WritableStreamDefaultController class in something close to the syntax of [ECMASCRIPT], it would look like

class WritableStreamDefaultController {
  constructor() // always throws

  error(e)
}

4.7.2. Internal Slots

Instances of WritableStreamDefaultController are created with the internal slots described in the following table:

Internal Slot Description (non-normative)
[[abortAlgorithm]] A promise-returning algorithm, taking one argument (the abort reason), which communicates a requested abort to the underlying sink
[[closeAlgorithm]] A promise-returning algorithm which communicates a requested close to the underlying sink
[[controlledWritableStream]] The WritableStream instance controlled
[[queue]] A List representing the stream’s internal queue of chunks
[[queueTotalSize]] The total size of all the chunks stored in [[queue]] (see §6.3 Queue-with-Sizes Operations)
[[started]] A boolean flag indicating whether the underlying sink has finished starting
[[strategyHWM]] A number supplied by the creator of the stream as part of the stream’s queuing strategy, indicating the point at which the stream will apply backpressure to its underlying sink
[[strategySizeAlgorithm]] An algorithm to calculate the size of enqueued chunks, as part of the stream’s queuing strategy
[[writeAlgorithm]] A promise-returning algorithm, taking one argument (the chunk to write), which writes data to the underlying sink

4.7.3. new WritableStreamDefaultController()

The WritableStreamDefaultController constructor cannot be used directly; WritableStreamDefaultController instances are created automatically during WritableStream construction.
  1. Throw a TypeError exception.

4.7.4. Properties of the WritableStreamDefaultController Prototype

4.7.4.1. error(e)
The error method will error the writable stream, making all future interactions with it fail with the given error e.

This method is rarely used, since usually it suffices to return a rejected promise from one of the underlying sink’s methods. However, it can be useful for suddenly shutting down a stream in response to an event outside the normal lifecycle of interactions with the underlying sink.

  1. If ! IsWritableStreamDefaultController(this) is false, throw a TypeError exception.
  2. Let state be this.[[controlledWritableStream]].[[state]].
  3. If state is not "writable", return.
  4. Perform ! WritableStreamDefaultControllerError(this, e).

4.7.5. Writable Stream Default Controller Internal Methods

The following are additional internal methods implemented by each WritableStreamDefaultController instance. The writable stream implementation will call into these.

The reason these are in method form, instead of as abstract operations, is to make it clear that the writable stream implementation is decoupled from the controller implementation, and could in the future be expanded with other controllers, as long as those controllers implemented such internal methods. A similar scenario is seen for readable streams, where there actually are multiple controller types and as such the counterpart internal methods are used polymorphically.

4.7.5.1. [[AbortSteps]]( reason )
  1. Return the result of performing this.[[abortAlgorithm]], passing reason.
4.7.5.2. [[ErrorSteps]]()
  1. Perform ! ResetQueue(this).

4.8. Writable Stream Default Controller Abstract Operations

4.8.1. IsWritableStreamDefaultController ( x ) nothrow

  1. If Type(x) is not Object, return false.
  2. If x does not have an [[controlledWritableStream]] internal slot, return false.
  3. Return true.

4.8.2. SetUpWritableStreamDefaultController ( stream, controller, startAlgorithm, writeAlgorithm, closeAlgorithm, abortAlgorithm, highWaterMark, sizeAlgorithm ) throws

  1. Assert: ! IsWritableStream(stream) is true.
  2. Assert: stream.[[writableStreamController]] is undefined.
  3. Set controller.[[controlledWritableStream]] to stream.
  4. Set stream.[[writableStreamController]] to controller.
  5. Perform ! ResetQueue(controller).
  6. Set controller.[[started]] to false.
  7. Set controller.[[strategySizeAlgorithm]] to sizeAlgorithm.
  8. Set controller.[[strategyHWM]] to highWaterMark.
  9. Set controller.[[writeAlgorithm]] to writeAlgorithm.
  10. Set controller.[[closeAlgorithm]] to closeAlgorithm.
  11. Set controller.[[abortAlgorithm]] to abortAlgorithm.
  12. Let backpressure be ! WritableStreamDefaultControllerGetBackpressure(controller).
  13. Perform ! WritableStreamUpdateBackpressure(stream, backpressure).
  14. Let startResult be the result of performing startAlgorithm. (This may throw an exception.)
  15. Let startPromise be a promise resolved with startResult.
  16. Upon fulfillment of startPromise,
    1. Assert: stream.[[state]] is "writable" or "erroring".
    2. Set controller.[[started]] to true.
    3. Perform ! WritableStreamDefaultControllerAdvanceQueueIfNeeded(controller).
  17. Upon rejection of startPromise with reason r,
    1. Assert: stream.[[state]] is "writable" or "erroring".
    2. Set controller.[[started]] to true.
    3. Perform ! WritableStreamDealWithRejection(stream, r).

4.8.3. SetUpWritableStreamDefaultControllerFromUnderlyingSink ( stream, underlyingSink, highWaterMark, sizeAlgorithm ) throws

  1. Assert: underlyingSink is not undefined.
  2. Let controller be ObjectCreate(the original value of WritableStreamDefaultController's prototype property).
  3. Let startAlgorithm be the following steps:
    1. Return ? InvokeOrNoop(underlyingSink, "start", « controller »).
  4. Let writeAlgorithm be ? CreateAlgorithmFromUnderlyingMethod(underlyingSink, "write", 1, « controller »).
  5. Let closeAlgorithm be ? CreateAlgorithmFromUnderlyingMethod(underlyingSink, "close", 0, « »).
  6. Let abortAlgorithm be ? CreateAlgorithmFromUnderlyingMethod(underlyingSink, "abort", 1, « »).
  7. Perform ? SetUpWritableStreamDefaultController(stream, controller, startAlgorithm, writeAlgorithm, closeAlgorithm, abortAlgorithm, highWaterMark, sizeAlgorithm).

4.8.4. WritableStreamDefaultControllerClose ( controller ) nothrow

  1. Perform ! EnqueueValueWithSize(controller, "close", 0).
  2. Perform ! WritableStreamDefaultControllerAdvanceQueueIfNeeded(controller).

4.8.5. WritableStreamDefaultControllerGetChunkSize ( controller, chunk ) nothrow

  1. Let returnValue be the result of performing controller.[[strategySizeAlgorithm]], passing in chunk, and interpreting the result as an ECMAScript completion value.
  2. If returnValue is an abrupt completion,
    1. Perform ! WritableStreamDefaultControllerErrorIfNeeded(controller, returnValue.[[Value]]).
    2. Return 1.
  3. Return returnValue.[[Value]].

4.8.6. WritableStreamDefaultControllerGetDesiredSize ( controller ) nothrow

  1. Return controller.[[strategyHWM]] − controller.[[queueTotalSize]].

4.8.7. WritableStreamDefaultControllerWrite ( controller, chunk, chunkSize ) nothrow

  1. Let writeRecord be Record {[[chunk]]: chunk}.
  2. Let enqueueResult be EnqueueValueWithSize(controller, writeRecord, chunkSize).
  3. If enqueueResult is an abrupt completion,
    1. Perform ! WritableStreamDefaultControllerErrorIfNeeded(controller, enqueueResult.[[Value]]).
    2. Return.
  4. Let stream be controller.[[controlledWritableStream]].
  5. If ! WritableStreamCloseQueuedOrInFlight(stream) is false and stream.[[state]] is "writable",
    1. Let backpressure be ! WritableStreamDefaultControllerGetBackpressure(controller).
    2. Perform ! WritableStreamUpdateBackpressure(stream, backpressure).
  6. Perform ! WritableStreamDefaultControllerAdvanceQueueIfNeeded(controller).

4.8.8. WritableStreamDefaultControllerAdvanceQueueIfNeeded ( controller ) nothrow

  1. Let stream be controller.[[controlledWritableStream]].
  2. If controller.[[started]] is false, return.
  3. If stream.[[inFlightWriteRequest]] is not undefined, return.
  4. Let state be stream.[[state]].
  5. If state is "closed" or "errored", return.
  6. If state is "erroring",
    1. Perform ! WritableStreamFinishErroring(stream).
    2. Return.
  7. If controller.[[queue]] is empty, return.
  8. Let writeRecord be ! PeekQueueValue(controller).
  9. If writeRecord is "close", perform ! WritableStreamDefaultControllerProcessClose(controller).
  10. Otherwise, perform ! WritableStreamDefaultControllerProcessWrite(controller, writeRecord.[[chunk]]).

4.8.9. WritableStreamDefaultControllerErrorIfNeeded ( controller, error ) nothrow

  1. If controller.[[controlledWritableStream]].[[state]] is "writable", perform ! WritableStreamDefaultControllerError(controller, error).

4.8.10. WritableStreamDefaultControllerProcessClose ( controller ) nothrow

  1. Let stream be controller.[[controlledWritableStream]].
  2. Perform ! WritableStreamMarkCloseRequestInFlight(stream).
  3. Perform ! DequeueValue(controller).
  4. Assert: controller.[[queue]] is empty.
  5. Let sinkClosePromise be the result of performing controller.[[closeAlgorithm]].
  6. Upon fulfillment of sinkClosePromise,
    1. Perform ! WritableStreamFinishInFlightClose(stream).
  7. Upon rejection of sinkClosePromise with reason reason,
    1. Perform ! WritableStreamFinishInFlightCloseWithError(stream, reason).

4.8.11. WritableStreamDefaultControllerProcessWrite ( controller, chunk ) nothrow

  1. Let stream be controller.[[controllerWritableStream]].
  2. Perform ! WritableStreamMarkFirstWriteRequestInFlight(stream).
  3. Let sinkWritePromise be the result of performing controller.[[writeAlgorithm]], passing in chunk.
  4. Upon fulfillment of sinkWritePromise,
    1. Perform ! WritableStreamFinishInFlightWrite(stream).
    2. Let state be stream.[[state]].
    3. Assert: state is "writable" or "erroring".
    4. Perform ! DequeueValue(controller).
    5. If ! WritableStreamCloseQueuedOrInFlight(stream) is false and state is "writable",
      1. Let backpressure be ! WritableStreamDefaultControllerGetBackpressure(controller).
      2. Perform ! WritableStreamUpdateBackpressure(stream, backpressure).
    6. Perform ! WritableStreamDefaultControllerAdvanceQueueIfNeeded(controller).
  5. Upon rejection of sinkWritePromise with reason,
    1. Perform ! WritableStreamFinishInFlightWriteWithError(stream, reason).

4.8.12. WritableStreamDefaultControllerGetBackpressure ( controller ) nothrow

  1. Let desiredSize be ! WritableStreamDefaultControllerGetDesiredSize(controller).
  2. Return desiredSize0.

4.8.13. WritableStreamDefaultControllerError ( controller, error ) nothrow

  1. Let stream be controller.[[controlledWritableStream]].
  2. Assert: stream.[[state]] is "writable".
  3. Perform ! WritableStreamStartErroring(stream, error).

5. Transform Streams

5.1. Using Transform Streams

The natural way to use a transform stream is to place it in a pipe between a readable stream and a writable stream. Chunks that travel from the readable stream to the writable stream will be transformed as they pass through the transform stream. Backpressure is respected, so data will not be read faster than it can be transformed and consumed.
readableStream
  .pipeThrough(transformStream)
  .pipeTo(writableStream)
  .then(() => console.log("All data successfully transformed!"))
  .catch(e => console.error("Something went wrong!", e));
You can also use the readable and writable properties of a transform stream directly to access the usual interfaces of a readable stream and writable stream. In this example we supply data to the writable side of the stream using its writer interface. The readable side is then piped to anotherWritableStream.
const writer = transformStream.writable.getWriter();
writer.write("input chunk");
transformStream.readable.pipeTo(anotherWritableStream);
One use of identity transform streams is to easily convert between readable and writable streams. For example, the fetch() API accepts a readable stream request body, but it can be more convenient to write data for uploading via a writable stream interface. Using an identity transform stream addresses this:
const { writable, readable } = new TransformStream();
fetch("...", { body: readable }).then(response => /* ... */);

writable.write(new Uint8Array([0x73, 0x74, 0x72, 0x65, 0x61, 0x6D, 0x73, 0x21]));
writable.close();

Another use of identity transform streams is to add additional buffering to a pipe. In this example we add extra buffering between readableStream and writableStream.

const writableStrategy = new ByteLengthQueuingStrategy({ highWaterMark: 1024 * 1024 });

readableStream
  .pipeThrough(new TransformStream(undefined, writableStrategy))
  .pipeTo(writableStream);

5.2. Class TransformStream

5.2.1. Class Definition

This section is non-normative.

If one were to write the TransformStream class in something close to the syntax of [ECMASCRIPT], it would look like

class TransformStream {
  constructor(transformer = {}, writableStrategy = {}, readableStrategy = {})

  get readable()
  get writable()
}

5.2.2. Internal Slots

Instances of TransformStream are created with the internal slots described in the following table:

Internal Slot Description (non-normative)
[[backpressure]] Whether there was backpressure on [[readable]] the last time it was observed
[[backpressureChangePromise]] A promise which is fulfilled and replaced every time the value of [[backpressure]] changes
[[readable]] The ReadableStream instance controlled by this object
[[transformStreamController]] A TransformStreamDefaultController created with the ability to control [[readable]] and [[writable]]; also used for the IsTransformStream brand check
[[writable]] The WritableStream instance controlled by this object

5.2.3. new TransformStream(transformer = {}, writableStrategy = {}, readableStrategy = {})

The transformer object passed to the constructor can implement any of the following methods to govern how the constructed stream instance behaves:

The controller object passed to start, transform and flush is an instance of TransformStreamDefaultController, and has the ability to enqueue chunks to the readable side, or to terminate or error the stream.

The second and third arguments to the constructor are the queuing strategy objects for the writable and readable sides respectively. These are used in the construction of the WritableStream and ReadableStream objects and can be used to add buffering to a TransformStream, in order to smooth out variations in the speed of the transformation, or to increase the amount of buffering in a pipe.

  1. Let readableType be ? GetV(transformer, "readableType").
  2. If readableType is not undefined, throw a RangeError exception.
  3. Let writableType be ? GetV(transformer, "writableType").
  4. If writableType is not undefined, throw a RangeError exception.
  5. Let writableSizeFunction be ? GetV(writableStrategy, "size").
  6. Let writableSizeAlgorithm be ? MakeSizeAlgorithmFromSizeFunction(writableSizeFunction).
  7. Let writableHighWaterMark be ? GetV(writableStrategy, "highWaterMark").
  8. If writableHighWaterMark is undefined, set writableHighWaterMark to 1.
  9. Set writableHighWaterMark to ? ValidateAndNormalizeHighWaterMark(writableHighWaterMark).
  10. Let readableSizeFunction be ? GetV(readableStrategy, "size").
  11. Let readableSizeAlgorithm be ? MakeSizeAlgorithmFromSizeFunction(readableSizeFunction).
  12. Let readableHighWaterMark be ? GetV(readableStrategy, "highWaterMark").
  13. If readableHighWaterMark is undefined, set readableHighWaterMark to 0.
  14. Set readableHighWaterMark be ? ValidateAndNormalizeHighWaterMark(readableHighWaterMark).
  15. Let startPromise be a new promise.
  16. Perform ! InitializeTransformStream(this, startPromise, writableHighWaterMark, writableSizeAlgorithm, readableHighWaterMark, readableSizeAlgorithm).
  17. Perform ? SetUpTransformStreamDefaultControllerFromTransformer(this, transformer).
  18. Let startResult be ? InvokeOrNoop(transformer, "start", « this.[[transformStreamController]] »).
  19. Resolve startPromise with startResult.

5.2.4. Properties of the TransformStream Prototype

5.2.4.1. get readable
The readable getter gives access to the readable side of the transform stream.
  1. If ! IsTransformStream(this) is false, throw a TypeError exception.
  2. Return this.[[readable]].
5.2.4.2. get writable
The writable getter gives access to the writable side of the transform stream.
  1. If ! IsTransformStream(this) is false, throw a TypeError exception.
  2. Return this.[[writable]].

5.3. General Transform Stream Abstract Operations

5.3.1. CreateTransformStream ( startAlgorithm, transformAlgorithm, flushAlgorithm [, writableHighWaterMark [, writableSizeAlgorithm [, readableHighWaterMark [, readableSizeAlgorithm ] ] ] ] ) throws

This abstract operation is meant to be called from other specifications that wish to create TransformStream instances. The transformAlgorithm and flushAlgorithm algorithms must return promises; if supplied, writableHighWaterMark and readableHighWaterMark must be nonnegative, non-NaN numbers; and if supplied, writableSizeAlgorithm and readableSizeAlgorithm must be algorithms accepting chunk objects and returning numbers.

CreateTransformStream throws an exception if and only if the supplied startAlgorithm throws.

  1. If writableHighWaterMark was not passed, set it to 1.
  2. If writableSizeAlgorithm was not passed, set it to an algorithm that returns 1.
  3. If readableHighWaterMark was not passed, set it to 0.
  4. If readableSizeAlgorithm was not passed, set it to an algorithm that returns 1.
  5. Assert: ! IsNonNegativeNumber(writableHighWaterMark) is true.
  6. Assert: ! IsNonNegativeNumber(readableHighWaterMark) is true.
  7. Let stream be ObjectCreate(the original value of TransformStream's prototype property).
  8. Let startPromise be a new promise.
  9. Perform ! InitializeTransformStream(stream, startPromise, writableHighWaterMark, writableSizeAlgorithm, readableHighWaterMark, readableSizeAlgorithm).
  10. Let controller be ObjectCreate(the original value of TransformStreamDefaultController's prototype property).
  11. Perform ! SetUpTransformStreamDefaultController(stream, controller, transformAlgorithm, flushAlgorithm).
  12. Let startResult be the result of performing startAlgorithm. (This may throw an exception.)
  13. Resolve startPromise with startResult.
  14. Return stream.

5.3.2. InitializeTransformStream ( stream, startPromise, writableHighWaterMark, writableSizeAlgorithm, readableHighWaterMark, readableSizeAlgorithm ) nothrow

  1. Let startAlgorithm be an algorithm that returns startPromise.
  2. Let writeAlgorithm be the following steps, taking a chunk argument:
    1. Return ! TransformStreamDefaultSinkWriteAlgorithm(stream, chunk).
  3. Let abortAlgorithm be the following steps:
    1. Return ! TransformStreamDefaultSinkAbortAlgorithm(stream).
  4. Let closeAlgorithm be the following steps:
    1. Return ! TransformStreamDefaultSinkCloseAlgorithm(stream).
  5. Set stream.[[writable]] to ! CreateWritableStream(startAlgorithm, writeAlgorithm, closeAlgorithm, abortAlgorithm, writableHighWaterMark, writableSizeAlgorithm).
  6. Let pullAlgorithm be the following steps:
    1. Return ! TransformStreamDefaultSourcePullAlgorithm(stream).
  7. Let cancelAlgorithm be the following steps, taking a reason argument:
    1. Perform ! TransformStreamErrorWritableAndUnblockWrite(stream, reason).
    2. Return a promise resolved with undefined.
  8. Set stream.[[readable]] to ! CreateReadableStream(startAlgorithm, pullAlgorithm, cancelAlgorithm, readableHighWaterMark, readableSizeAlgorithm).
  9. Set stream.[[backpressure]] and stream.[[backpressureChangePromise]] to undefined.

    The [[backpressure]] slot is set to undefined so that it can be initialized by TransformStreamSetBackpressure. Alternatively, implementations can use a strictly boolean value for [[backpressure]] and change the way it is initialized. This will not be visible to user code so long as the initialization is correctly completed before transformer’s start() method is is called.

  10. Perform ! TransformStreamSetBackpressure(stream, true).
  11. Set stream.[[transformStreamController]] to undefined.

5.3.3. IsTransformStream ( x ) nothrow

  1. If Type(x) is not Object, return false.
  2. If x does not have a [[transformStreamController]] internal slot, return false.
  3. Return true.

5.3.4. TransformStreamError ( stream, e ) nothrow

  1. If stream.[[readable]].[[state]] is "readable", perform ! ReadableStreamDefaultControllerError(stream.[[readable]].[[readableStreamController]], e).
  2. Perform ! TransformStreamErrorWritableAndUnblockWrite(stream, e).

This operation works correctly when one or both sides are already errored. As a result, calling algorithms do not need to check stream states when responding to an error condition.

5.3.5. TransformStreamErrorWritableAndUnblockWrite ( stream, e ) nothrow

  1. Perform ! WritableStreamDefaultControllerErrorIfNeeded(stream.[[writable]].[[writableStreamController]], e).
  2. If stream.[[backpressure]] is true, perform ! TransformStreamSetBackpressure(stream, false).

The TransformStreamDefaultSinkWriteAlgorithm abstract operation could be waiting for the promise stored in the [[backpressureChangePromise]] slot to resolve. This call to TransformStreamSetBackpressure ensures that the promise always resolves.

5.3.6. TransformStreamSetBackpressure ( stream, backpressure ) nothrow

  1. Assert: stream.[[backpressure]] is not backpressure.
  2. If stream.[[backpressureChangePromise]] is not undefined, resolve stream.[[backpressureChangePromise]] with undefined.
  3. Set stream.[[backpressureChangePromise]] to a new promise.
  4. Set stream.[[backpressure]] to backpressure.

5.4. Class TransformStreamDefaultController

The TransformStreamDefaultController class has methods that allow manipulation of the associated ReadableStream and WritableStream. When constructing a TransformStream, the transformer object is given a corresponding TransformStreamDefaultController instance to manipulate.

5.4.1. Class Definition

This section is non-normative.

If one were to write the TransformStreamDefaultController class in something close to the syntax of [ECMASCRIPT], it would look like

class TransformStreamDefaultController {
  constructor() // always throws

  get desiredSize()

  enqueue(chunk)
  error(reason)
  terminate()
}

5.4.2. Internal Slots

Instances of TransformStreamDefaultController are created with the internal slots described in the following table:

Internal Slot Description (non-normative)
[[controlledTransformStream]] The TransformStream instance controlled; also used for the IsTransformStreamDefaultController brand check
[[flushAlgorithm]] A promise-returning algorithm which communicates a requested close to the transformer
[[transformAlgorithm]] A promise-returning algorithm, taking one argument (the chunk to transform), which requests the transformer perform its transformation

5.4.3. new TransformStreamDefaultController()

The TransformStreamDefaultController constructor cannot be used directly; TransformStreamDefaultController instances are created automatically during TransformStream construction.
  1. Throw a TypeError exception.

5.4.4. Properties of the TransformStreamDefaultController Prototype

5.4.4.1. get desiredSize
The desiredSize getter returns the desired size to fill the readable side’s internal queue. It can be negative, if the queue is over-full.
  1. If ! IsTransformStreamDefaultController(this) is false, throw a TypeError exception.
  2. Let readableController be this.[[controlledTransformStream]].[[readable]].[[readableStreamController]].
  3. Return ! ReadableStreamDefaultControllerGetDesiredSize(readableController).
5.4.4.2. enqueue(chunk)
The enqueue method will enqueue a given chunk in the readable side.
  1. If ! IsTransformStreamDefaultController(this) is false, throw a TypeError exception.
  2. Perform ? TransformStreamDefaultControllerEnqueue(this, chunk).
5.4.4.3. error(reason)
The error method will error both the readable side and the writable side of the controlled transform stream, making all future interactions fail with the given reason. Any chunks queued for transformation will be discarded.
  1. If ! IsTransformStreamDefaultController(this) is false, throw a TypeError exception.
  2. Perform ! TransformStreamDefaultControllerError(this, reason).
5.4.4.4. terminate()
The terminate method will close the readable side and error the writable side of the controlled transform stream. This is useful when the transformer only needs to consume a portion of the chunks written to the writable side.
  1. If ! IsTransformStreamDefaultController(this) is false, throw a TypeError exception.
  2. Perform ! TransformStreamDefaultControllerTerminate(this).

5.5. Transform Stream Default Controller Abstract Operations

5.5.1. IsTransformStreamDefaultController ( x ) nothrow

  1. If Type(x) is not Object, return false.
  2. If x does not have an [[controlledTransformStream]] internal slot, return false.
  3. Return true.

5.5.2. SetUpTransformStreamDefaultController ( stream, controller, transformAlgorithm, flushAlgorithm ) nothrow

  1. Assert: ! IsTransformStream(stream) is true.
  2. Assert: stream.[[writableStreamController]] is undefined.
  3. Set controller.[[controlledTransformStream]] to stream.
  4. Set stream.[[transformStreamController]] to controller.
  5. Set controller.[[transformAlgorithm]] to transformAlgorithm.
  6. Set controller.[[flushAlgorithm]] to flushAlgorithm.

5.5.3. SetUpTransformStreamDefaultControllerFromTransformer ( stream, transformer ) throws

  1. Assert: transformer is not undefined.
  2. Let controller be ObjectCreate(the original value of TransformStreamDefaultController's prototype property).
  3. Let transformAlgorithm be the following steps, taking a chunk argument:
    1. Let result be TransformStreamDefaultControllerEnqueue(controller, chunk).
    2. If result is an abrupt completion, return a promise rejected with result.[[Value]].
    3. Otherwise, return a promise resolved with undefined.
  4. Let transformMethod be ? GetV(transformer, "transform").
  5. If transformMethod is not undefined,
    1. If ! IsCallable(transformMethod) is false, throw a TypeError exception.
    2. Set transformAlgorithm to the following steps, taking a chunk argument:
      1. Let transformPromise be ! PromiseCall(transformMethod, transformer, « chunk, controller »).
      2. Return the result of transforming transformPromise with a rejection handler that, when called with argument e, performs the following steps:
        1. Perform ! TransformStreamError(stream, e).
        2. Throw e.
  6. Let flushAlgorithm be ? CreateAlgorithmFromUnderlyingMethod(transformer, "flush", 0, « controller »).
  7. Perform ! SetUpTransformStreamDefaultController(stream, controller, transformAlgorithm, flushAlgorithm).

5.5.4. TransformStreamDefaultControllerEnqueue ( controller, chunk ) throws

This abstract operation can be called by other specifications that wish to enqueue chunks in the readable side, in the same way a developer would enqueue chunks using the stream’s associated controller object. Specifications should not do this to streams they did not create.

  1. Let stream be controller.[[controlledTransformStream]].
  2. Let readableController be stream.[[readable]].[[readableStreamController]].
  3. If ! ReadableStreamDefaultControllerCanCloseOrEnqueue(readableController) is false, throw a TypeError exception.
  4. Let enqueueResult be ReadableStreamDefaultControllerEnqueue(readableController, chunk).
  5. If enqueueResult is an abrupt completion,
    1. Perform ! TransformStreamErrorWritableAndUnblockWrite(stream, enqueueResult.[[Value]]).
    2. Throw stream.[[readable]].[[storedError]].
  6. Let backpressure be ! ReadableStreamDefaultControllerHasBackpressure(readableController).
  7. If backpressure is not stream.[[backpressure]],
    1. Assert: backpressure is true.
    2. Perform ! TransformStreamSetBackpressure(stream, true).

5.5.5. TransformStreamDefaultControllerError ( controller, e ) nothrow

This abstract operation can be called by other specifications that wish to move a transform stream to an errored state, in the same way a developer would error a stream using its associated controller object. Specifications should not do this to streams they did not create.

  1. Perform ! TransformStreamError(controller.[[controlledTransformStream]], e).

5.5.6. TransformStreamDefaultControllerTerminate ( controller ) nothrow

This abstract operation can be called by other specifications that wish to terminate a transform stream, in the same way a developer-created stream would be closed by its associated controller object. Specifications should not do this to streams they did not create.

  1. Let stream be controller.[[controlledTransformStream]].
  2. Let readableController be stream.[[readable]].[[readableStreamController]].
  3. If ! ReadableStreamDefaultControllerCanCloseOrEnqueue(readableController) is true, perform ! ReadableStreamDefaultControllerClose(readableController).
  4. Let error be a TypeError exception indicating that the stream has been terminated.
  5. Perform ! TransformStreamErrorWritableAndUnblockWrite(stream, error).

5.6. Transform Stream Default Sink Abstract Operations

5.6.1. TransformStreamDefaultSinkWriteAlgorithm ( stream, chunk ) nothrow

  1. Assert: stream.[[writable]].[[state]] is "writable".
  2. Let controller be stream.[[transformStreamController]].
  3. If stream.[[backpressure]] is true,
    1. Let backpressureChangePromise be stream.[[backpressureChangePromise]].
    2. Assert: backpressureChangePromise is not undefined.
    3. Return the result of transforming backpressureChangePromise with a fulfillment handler which performs the following steps:
      1. Let writable be stream.[[writable]].
      2. Let state be writable.[[state]].
      3. If state is "erroring", throw writable.[[storedError]].
      4. Assert: state is "writable".
      5. Return the result of performing controller.[[transformAlgorithm]], passing chunk.
  4. Return the result of performing controller.[[transformAlgorithm]], passing chunk.

5.6.2. TransformStreamDefaultSinkAbortAlgorithm ( stream ) nothrow

  1. Let e be a TypeError exception indicating that the writable stream was aborted.
  2. Perform ! TransformStreamError(stream, e).
  3. Return a promise resolved with undefined.

5.6.3. TransformStreamDefaultSinkCloseAlgorithm( stream ) nothrow

  1. Let readable be stream.[[readable]].
  2. Let flushPromise be the result of performing stream.[[transformStreamController]].[[flushAlgorithm]].
  3. Return the result of transforming flushPromise with:
    1. A fulfullment handler that performs the following steps:
      1. If readable.[[state]] is "errored", throw readable.[[storedError]].
      2. Let readableController be readable.[[readableStreamController]].
      3. If ! ReadableStreamDefaultControllerCanCloseOrEnqueue(readableController) is true, perform ! ReadableStreamDefaultControllerClose(readableController).
    2. A rejection handler that, when called with argument r, performs the following steps:
      1. Perform ! TransformStreamError(stream, r).
      2. Throw readable.[[storedError]].

5.7. Transform Stream Default Source Abstract Operations

5.7.1. TransformStreamDefaultSourcePullAlgorithm( stream ) nothrow

  1. Assert: stream.[[backpressure]] is true.
  2. Assert: stream.[[backpressureChangePromise]] is not undefined.
  3. Perform ! TransformStreamSetBackpressure(stream, false).
  4. Return stream.[[backpressureChangePromise]].

6. Other Stream APIs and Operations

6.1. Class ByteLengthQueuingStrategy

A common queuing strategy when dealing with bytes is to wait until the accumulated byteLength properties of the incoming chunks reaches a specified high-water mark. As such, this is provided as a built-in queuing strategy that can be used when constructing streams.

When creating a readable stream or writable stream, you can supply a byte-length queuing strategy directly:
const stream = new ReadableStream(
  { ... },
  new ByteLengthQueuingStrategy({ highWaterMark: 16 * 1024 })
);

In this case, 16 KiB worth of chunks can be enqueued by the readable stream’s underlying source before the readable stream implementation starts sending backpressure signals to the underlying source.

const stream = new WritableStream(
  { ... },
  new ByteLengthQueuingStrategy({ highWaterMark: 32 * 1024 })
);

In this case, 32 KiB worth of chunks can be accumulated in the writable stream’s internal queue, waiting for previous writes to the underlying sink to finish, before the writable stream starts sending backpressure signals to any producers.

It is not necessary to use ByteLengthQueuingStrategy with readable byte streams, as they always measure chunks in bytes. Attempting to construct a byte stream with a ByteLengthQueuingStrategy will fail.

6.1.1. Class Definition

This section is non-normative.

If one were to write the ByteLengthQueuingStrategy class in something close to the syntax of [ECMASCRIPT], it would look like

class ByteLengthQueuingStrategy {
  constructor({ highWaterMark })
  size(chunk)
}

Each ByteLengthQueuingStrategy instance will additionally have an own data property highWaterMark set by its constructor.

6.1.2. new ByteLengthQueuingStrategy({ highWaterMark })

The constructor takes a non-negative number for the high-water mark, and stores it on the object as a property.
  1. Perform ! CreateDataProperty(this, "highWaterMark", highWaterMark).

6.1.3. Properties of the ByteLengthQueuingStrategy Prototype

6.1.3.1. size(chunk)
The size method returns the given chunk’s byteLength property. (If the chunk doesn’t have one, it will return undefined, causing the stream using this strategy to error.)

This method is intentionally generic; it does not require that its this value be a ByteLengthQueuingStrategy object.

  1. Return ? GetV(chunk, "byteLength").

6.2. Class CountQueuingStrategy

A common queuing strategy when dealing with streams of generic objects is to simply count the number of chunks that have been accumulated so far, waiting until this number reaches a specified high-water mark. As such, this strategy is also provided out of the box.

When creating a readable stream or writable stream, you can supply a count queuing strategy directly:
const stream = new ReadableStream(
  { ... },
  new CountQueuingStrategy({ highWaterMark: 10 })
);

In this case, 10 chunks (of any kind) can be enqueued by the readable stream’s underlying source before the readable stream implementation starts sending backpressure signals to the underlying source.

const stream = new WritableStream(
  { ... },
  new CountQueuingStrategy({ highWaterMark: 5 })
);

In this case, five chunks (of any kind) can be accumulated in the writable stream’s internal queue, waiting for previous writes to the underlying sink to finish, before the writable stream starts sending backpressure signals to any producers.

6.2.1. Class Definition

This section is non-normative.

If one were to write the CountQueuingStrategy class in something close to the syntax of [ECMASCRIPT], it would look like

class CountQueuingStrategy {
  constructor({ highWaterMark })
  size()
}

Each CountQueuingStrategy instance will additionally have an own data property highWaterMark set by its constructor.

6.2.2. new CountQueuingStrategy({ highWaterMark })

The constructor takes a non-negative number for the high-water mark, and stores it on the object as a property.
  1. Perform ! CreateDataProperty(this, "highWaterMark", highWaterMark).

6.2.3. Properties of the CountQueuingStrategy Prototype

6.2.3.1. size()
The size method returns one always, so that the total queue size is a count of the number of chunks in the queue.

This method is intentionally generic; it does not require that its this value be a CountQueuingStrategy object.

  1. Return 1.

6.3. Queue-with-Sizes Operations

The streams in this specification use a "queue-with-sizes" data structure to store queued up values, along with their determined sizes. Various specification objects contain a queue-with-sizes, represented by the object having two paired internal slots, always named [[queue]] and [[queueTotalSize]]. [[queue]] is a List of Records with [[value]] and [[size]] fields, and [[queueTotalSize]] is a JavaScript Number, i.e. a double-precision floating point number.

The following abstract operations are used when operating on objects that contain queues-with-sizes, in order to ensure that the two internal slots stay synchronized.

Due to the limited precision of floating-point arithmetic, the framework specified here, of keeping a running total in the [[queueTotalSize]] slot, is not equivalent to adding up the size of all chunks in [[queue]]. (However, this only makes a difference when there is a huge (~1015) variance in size between chunks, or when trillions of chunks are enqueued.)

6.3.1. DequeueValue ( container ) nothrow

  1. Assert: container has [[queue]] and [[queueTotalSize]] internal slots.
  2. Assert: container.[[queue]] is not empty.
  3. Let pair be the first element of container.[[queue]].
  4. Remove pair from container.[[queue]], shifting all other elements downward (so that the second becomes the first, and so on).
  5. Set container.[[queueTotalSize]] to container.[[queueTotalSize]] − pair.[[size]].
  6. If container.[[queueTotalSize]] < 0, set container.[[queueTotalSize]] to 0. (This can occur due to rounding errors.)
  7. Return pair.[[value]].

6.3.2. EnqueueValueWithSize ( container, value, size ) throws

  1. Assert: container has [[queue]] and [[queueTotalSize]] internal slots.
  2. Let size be ? ToNumber(size).
  3. If ! IsFiniteNonNegativeNumber(size) is false, throw a RangeError exception.
  4. Append Record {[[value]]: value, [[size]]: size} as the last element of container.[[queue]].
  5. Set container.[[queueTotalSize]] to container.[[queueTotalSize]] + size.

6.3.3. PeekQueueValue ( container ) nothrow

  1. Assert: container has [[queue]] and [[queueTotalSize]] internal slots.
  2. Assert: container.[[queue]] is not empty.
  3. Let pair be the first element of container.[[queue]].
  4. Return pair.[[value]].

6.3.4. ResetQueue ( container ) nothrow

  1. Assert: container has [[queue]] and [[queueTotalSize]] internal slots.
  2. Set container.[[queue]] to a new empty List.
  3. Set container.[[queueTotalSize]] to 0.

6.4. Miscellaneous Operations

A few abstract operations are used in this specification for utility purposes. We define them here.

6.4.1. CreateAlgorithmFromUnderlyingMethod ( underlyingObject, methodName, algoArgCount, extraArgs ) throws

  1. Assert: underlyingObject is not undefined.
  2. Assert: ! IsPropertyKey(methodName) is true.
  3. Assert: algoArgCount is 0 or 1.
  4. Assert: extraArgs is a List.
  5. Let method be ? GetV(underlyingObject, methodName).
  6. If method is not undefined,
    1. If ! IsCallable(method) is false, throw a TypeError exception.
    2. If algoArgCount is 0, return an algorithm that performs the following steps:
      1. Return ! PromiseCall(method, underlyingObject, extraArgs).
    3. Otherwise, return an algorithm that performs the following steps, taking an arg argument:
      1. Let fullArgs be a List consisting of arg followed by the elements of extraArgs in order.
      2. Return ! PromiseCall(method, underlyingObject, fullArgs).
  7. Return an algorithm which returns a promise resolved with undefined.

6.4.2. InvokeOrNoop ( O, P, args ) throws

InvokeOrNoop is a slight modification of the [ECMASCRIPT] Invoke abstract operation to return undefined when the method is not present.
  1. Assert: O is not undefined.
  2. Assert: ! IsPropertyKey(P) is true.
  3. Assert: args is a List.
  4. Let method be ? GetV(O, P).
  5. If method is undefined, return undefined.
  6. Return ? Call(method, O, args).

6.4.3. IsFiniteNonNegativeNumber ( v ) nothrow

  1. If ! IsNonNegativeNumber(v) is false, return false.
  2. If v is +∞, return false.
  3. Return true.

6.4.4. IsNonNegativeNumber ( v ) nothrow

  1. If Type(v) is not Number, return false.
  2. If v is NaN, return false.
  3. If v < 0, return false.
  4. Return true.

6.4.5. PromiseCall ( F, V, args ) nothrow

PromiseCall is a variant of promise-calling that works on methods.
  1. Assert: ! IsCallable(F) is true.
  2. Assert: V is not undefined.
  3. Assert: args is a List.
  4. Let returnValue be Call(F, V, args).
  5. If returnValue is an abrupt completion, return a promise rejected with returnValue.[[Value]].
  6. Otherwise, return a promise resolved with returnValue.[[Value]].

6.4.6. TransferArrayBuffer ( O ) nothrow

  1. Assert: Type(O) is Object.
  2. Assert: O has an [[ArrayBufferData]] internal slot.
  3. Assert: ! IsDetachedBuffer(O) is false.
  4. Let arrayBufferData be O.[[ArrayBufferData]].
  5. Let arrayBufferByteLength be O.[[ArrayBufferByteLength]].
  6. Perform ! DetachArrayBuffer(O).
  7. Return a new ArrayBuffer object (created in the current Realm Record) whose [[ArrayBufferData]] internal slot value is arrayBufferData and whose [[ArrayBufferByteLength]] internal slot value is arrayBufferByteLength.

6.4.7. ValidateAndNormalizeHighWaterMark ( highWaterMark ) throws

  1. Set highWaterMark to ? ToNumber(highWaterMark).
  2. If highWaterMark is NaN or highWaterMark < 0, throw a RangeError exception.

    +∞ is explicitly allowed as a valid high water mark. It causes backpressure to never be applied.

  3. Return highWaterMark.

6.4.8. MakeSizeAlgorithmFromSizeFunction ( size ) throws

  1. If size is undefined, return an algorithm that returns 1.
  2. If ! IsCallable(size) is false, throw a TypeError exception.
  3. Return an algorithm that performs the following steps, taking a chunk argument:
    1. Return ? Call(size, undefined, « chunk »).

7. Global Properties

The following constructors must be exposed on the global object as data properties of the same name:

The attributes of these properties must be { [[Writable]]: true, [[Enumerable]]: false, [[Configurable]]: true }.

The ReadableStreamDefaultReader, ReadableStreamBYOBReader, ReadableStreamDefaultController, ReadableByteStreamController, WritableStreamDefaultWriter, WritableStreamDefaultController, and TransformStreamDefaultController classes are specifically not exposed, as they are not independently useful.

8. Examples of Creating Streams

This section, and all its subsections, are non-normative.

The previous examples throughout the standard have focused on how to use streams. Here we show how to create a stream, using the ReadableStream or WritableStream constructors.

8.1. A readable stream with an underlying push source (no backpressure support)

The following function creates readable streams that wrap WebSocket instances [HTML], which are push sources that do not support backpressure signals. It illustrates how, when adapting a push source, usually most of the work happens in the start function.

function makeReadableWebSocketStream(url, protocols) {
  const ws = new WebSocket(url, protocols);
  ws.binaryType = "arraybuffer";

  return new ReadableStream({
    start(controller) {
      ws.onmessage = event => controller.enqueue(event.data);
      ws.onclose = () => controller.close();
      ws.onerror = () => controller.error(new Error("The WebSocket errored!"));
    },

    cancel() {
      ws.close();
    }
  });
}

We can then use this function to create readable streams for a web socket, and pipe that stream to an arbitrary writable stream:

const webSocketStream = makeReadableWebSocketStream("wss://example.com:443/", "protocol");

webSocketStream.pipeTo(writableStream)
  .then(() => console.log("All data successfully written!"))
  .catch(e => console.error("Something went wrong!", e));

8.2. A readable stream with an underlying push source and backpressure support

The following function returns readable streams that wrap "backpressure sockets," which are hypothetical objects that have the same API as web sockets, but also provide the ability to pause and resume the flow of data with their readStop and readStart methods. In doing so, this example shows how to apply backpressure to underlying sources that support it.

function makeReadableBackpressureSocketStream(host, port) {
  const socket = createBackpressureSocket(host, port);

  return new ReadableStream({
    start(controller) {
      socket.ondata = event => {
        controller.enqueue(event.data);

        if (controller.desiredSize <= 0) {
          // The internal queue is full, so propagate
          // the backpressure signal to the underlying source.
          socket.readStop();
        }
      };

      socket.onend = () => controller.close();
      socket.onerror = () => controller.error(new Error("The socket errored!"));
    },

    pull() {
      // This is called if the internal queue has been emptied, but the
      // stream’s consumer still wants more data. In that case, restart
      // the flow of data if we have previously paused it.
      socket.readStart();
    },

    cancel() {
      socket.close();
    }
  });
}

We can then use this function to create readable streams for such "backpressure sockets" in the same way we do for web sockets. This time, however, when we pipe to a destination that cannot accept data as fast as the socket is producing it, or if we leave the stream alone without reading from it for some time, a backpressure signal will be sent to the socket.

8.3. A readable byte stream with an underlying push source (no backpressure support)

The following function returns readable byte streams that wraps a hypothetical UDP socket API, including a promise-returning select2() method that is meant to be evocative of the POSIX select(2) system call.

Since the UDP protocol does not have any built-in backpressure support, the backpressure signal given by desiredSize is ignored, and the stream ensures that when data is available from the socket but not yet requested by the developer, it is enqueued in the stream’s internal queue, to avoid overflow of the kernel-space queue and a consequent loss of data.

This has some interesting consequences for how consumers interact with the stream. If the consumer does not read data as fast as the socket produces it, the chunks will remain in the stream’s internal queue indefinitely. In this case, using a BYOB reader will cause an extra copy, to move the data from the stream’s internal queue to the developer-supplied buffer. However, if the consumer consumes the data quickly enough, a BYOB reader will allow zero-copy reading directly into developer-supplied buffers.

(You can imagine a more complex version of this example which uses desiredSize to inform an out-of-band backpressure signaling mechanism, for example by sending a message down the socket to adjust the rate of data being sent. That is left as an exercise for the reader.)

const DEFAULT_CHUNK_SIZE = 65536;

function makeUDPSocketStream(host, port) {
  const socket = createUDPSocket(host, port);

  return new ReadableStream({
    type: "bytes",

    start(controller) {
      readRepeatedly().catch(e => controller.error(e));

      function readRepeatedly() {
        return socket.select2().then(() => {
          // Since the socket can become readable even when there’s
          // no pending BYOB requests, we need to handle both cases.
          let bytesRead;
          if (controller.byobRequest) {
            const v = controller.byobRequest.view;
            bytesRead = socket.readInto(v.buffer, v.byteOffset, v.byteLength);
            controller.byobRequest.respond(bytesRead);
          } else {
            const buffer = new ArrayBuffer(DEFAULT_CHUNK_SIZE);
            bytesRead = socket.readInto(buffer, 0, DEFAULT_CHUNK_SIZE);
            controller.enqueue(new Uint8Array(buffer, 0, bytesRead));
          }

          if (bytesRead === 0) {
            controller.close();
            return;
          }

          return readRepeatedly();
        });
      }
    },

    cancel() {
      socket.close();
    }
  });
}

ReadableStream instances returned from this function can now vend BYOB readers, with all of the aforementioned benefits and caveats.

8.4. A readable stream with an underlying pull source

The following function returns readable streams that wrap portions of the Node.js file system API (which themselves map fairly directly to C’s fopen, fread, and fclose trio). Files are a typical example of pull sources. Note how in contrast to the examples with push sources, most of the work here happens on-demand in the pull function, and not at startup time in the start function.

const fs = require("pr/fs"); // https://github.com/jden/pr
const CHUNK_SIZE = 1024;

function makeReadableFileStream(filename) {
  let fd;
  let position = 0;

  return new ReadableStream({
    start() {
      return fs.open(filename, "r").then(result => {
        fd = result;
      });
    },

    pull(controller) {
      const buffer = new ArrayBuffer(CHUNK_SIZE);

      return fs.read(fd, buffer, 0, CHUNK_SIZE, position).then(bytesRead => {
        if (bytesRead === 0) {
          return fs.close(fd).then(() => controller.close());
        } else {
          position += bytesRead;
          controller.enqueue(new Uint8Array(buffer, 0, bytesRead));
        }
      });
    },

    cancel() {
      return fs.close(fd);
    }
  });
}

We can then create and use readable streams for files just as we could before for sockets.

8.5. A readable byte stream with an underlying pull source

The following function returns readable byte streams that allow efficient zero-copy reading of files, again using the Node.js file system API. Instead of using a predetermined chunk size of 1024, it attempts to fill the developer-supplied buffer, allowing full control.

const fs = require("pr/fs"); // https://github.com/jden/pr
const DEFAULT_CHUNK_SIZE = 1024;

function makeReadableByteFileStream(filename) {
  let fd;
  let position = 0;

  return new ReadableStream({
    type: "bytes",

    start() {
      return fs.open(filename, "r").then(result => {
        fd = result;
      });
    },

    pull(controller) {
      // Even when the consumer is using the default reader, the auto-allocation
      // feature allocates a buffer and passes it to us via byobRequest.
      const v = controller.byobRequest.view;

      return fs.read(fd, v.buffer, v.byteOffset, v.byteLength, position).then(bytesRead => {
        if (bytesRead === 0) {
          return fs.close(fd).then(() => controller.close());
        } else {
          position += bytesRead;
          controller.byobRequest.respond(bytesRead);
        }
      });
    },

    cancel() {
      return fs.close(fd);
    },

    autoAllocateChunkSize: DEFAULT_CHUNK_SIZE
  });
}

With this in hand, we can create and use BYOB readers for the returned ReadableStream. But we can also create default readers, using them in the same simple and generic manner as usual. The adaptation between the low-level byte tracking of the underlying byte source shown here, and the higher-level chunk-based consumption of a default reader, is all taken care of automatically by the streams implementation. The auto-allocation feature, via the autoAllocateChunkSize option, even allows us to write less code, compared to the manual branching in §8.3 A readable byte stream with an underlying push source (no backpressure support).

8.6. A writable stream with no backpressure or success signals

The following function returns a writable stream that wraps a WebSocket [HTML]. Web sockets do not provide any way to tell when a given chunk of data has been successfully sent (without awkward polling of bufferedAmount, which we leave as an exercise to the reader). As such, this writable stream has no ability to communicate accurate backpressure signals or write success/failure to its producers. That is, the promises returned by its writer’s write() method and ready getter will always fulfill immediately.

function makeWritableWebSocketStream(url, protocols) {
  const ws = new WebSocket(url, protocols);

  return new WritableStream({
    start(controller) {
      ws.onerror = () => {
        controller.error(new Error("The WebSocket errored!"));
        ws.onclose = null;
      };
      ws.onclose = () => controller.error(new Error("The server closed the connection unexpectedly!"));
      return new Promise(resolve => ws.onopen = resolve);
    },

    write(chunk) {
      ws.send(chunk);
      // Return immediately, since the web socket gives us no easy way to tell
      // when the write completes.
    },

    close() {
      return closeWS(1000);
    },

    abort(reason) {
      return closeWS(4000, reason && reason.message);
    },
  });

  function closeWS(code, reasonString) {
    return new Promise((resolve, reject) => {
      ws.onclose = e => {
        if (e.wasClean) {
          resolve();
        } else {
          reject(new Error("The connection was not closed cleanly"));
        }
      };
      ws.close(code, reasonString);
    });
  }
}

We can then use this function to create writable streams for a web socket, and pipe an arbitrary readable stream to it:

const webSocketStream = makeWritableWebSocketStream("wss://example.com:443/", "protocol");

readableStream.pipeTo(webSocketStream)
  .then(() => console.log("All data successfully written!"))
  .catch(e => console.error("Something went wrong!", e));

8.7. A writable stream with backpressure and success signals

The following function returns writable streams that wrap portions of the Node.js file system API (which themselves map fairly directly to C’s fopen, fwrite, and fclose trio). Since the API we are wrapping provides a way to tell when a given write succeeds, this stream will be able to communicate backpressure signals as well as whether an individual write succeeded or failed.

const fs = require("pr/fs"); // https://github.com/jden/pr

function makeWritableFileStream(filename) {
  let fd;

  return new WritableStream({
    start() {
      return fs.open(filename, "w").then(result => {
        fd = result;
      });
    },

    write(chunk) {
      return fs.write(fd, chunk, 0, chunk.length);
    },

    close() {
      return fs.close(fd);
    },

    abort() {
      return fs.close(fd);
    }
  });
}

We can then use this function to create a writable stream for a file, and write individual chunks of data to it:

const fileStream = makeWritableFileStream("/example/path/on/fs.txt");
const writer = fileStream.getWriter();

writer.write("To stream, or not to stream\n");
writer.write("That is the question\n");

writer.close()
  .then(() => console.log("chunks written and stream closed successfully!"))
  .catch(e => console.error(e));

Note that if a particular call to fs.write takes a longer time, the returned promise will fulfill later. In the meantime, additional writes can be queued up, which are stored in the stream’s internal queue. The accumulation of chunks in this queue can change the stream to return a pending promise from the ready getter, which is a signal to producers that they would benefit from backing off and stopping writing, if possible.

The way in which the writable stream queues up writes is especially important in this case, since as stated in the documentation for fs.write, "it is unsafe to use fs.write multiple times on the same file without waiting for the [promise]." But we don’t have to worry about that when writing the makeWritableFileStream function, since the stream implementation guarantees that the underlying sink’s write method will not be called until any promises returned by previous calls have fulfilled!

8.8. A { readable, writable } stream pair wrapping the same underlying resource

The following function returns an object of the form { readable, writable }, with the readable property containing a readable stream and the writable property containing a writable stream, where both streams wrap the same underlying web socket resource. In essence, this combines §8.1 A readable stream with an underlying push source (no backpressure support) and §8.6 A writable stream with no backpressure or success signals.

While doing so, it illustrates how you can use JavaScript classes to create reusable underlying sink and underlying source abstractions.

function streamifyWebSocket(url, protocol) {
  const ws = new WebSocket(url, protocols);
  ws.binaryType = "arraybuffer";

  return {
    readable: new ReadableStream(new WebSocketSource(ws)),
    writable: new WritableStream(new WebSocketSink(ws))
  };
}

class WebSocketSource {
  constructor(ws) {
    this._ws = ws;
  }

  start(controller) {
    this._ws.onmessage = event => controller.enqueue(event.data);
    this._ws.onclose = () => controller.close();

    this._ws.addEventListener("error", () => {
      controller.error(new Error("The WebSocket errored!"));
    });
  }

  cancel() {
    this._ws.close();
  }
}

class WebSocketSink {
  constructor(ws) {
    this._ws = ws;
  }

  start(controller) {
    this._ws.onclose = () => controller.error(new Error("The server closed the connection unexpectedly!"));
    this._ws.addEventListener("error", () => {
      controller.error(new Error("The WebSocket errored!"));
      this._ws.onclose = null;
    });

    return new Promise(resolve => this._ws.onopen = resolve);
  }

  write(chunk) {
    this._ws.send(chunk);
  }

  close() {
    return this._closeWS(1000);
  }

  abort(reason) {
    return this._closeWS(4000, reason && reason.message);
  }

  _closeWS(code, reason) {
    return new Promise((resolve, reject) => {
      this._ws.onclose = e => {
        if (e.wasClean) {
          resolve();
        } else {
          reject(new Error("The connection was not closed cleanly"));
        }
      };
      this._ws.close(code, reasonString);
    });
  }
}

We can then use the objects created by this function to communicate with a remote web socket, using the standard stream APIs:

const streamyWS = streamifyWebSocket("wss://example.com:443/", "protocol");
const writer = streamyWS.writable.getWriter();
const reader = streamyWS.readable.getReader();

writer.write("Hello");
writer.write("web socket!");

reader.read().then(({ value, done }) => {
  console.log("The web socket says: ", value);
});

Note how in this setup canceling the readable side will implicitly close the writable side, and similarly, closing or aborting the writable side will implicitly close the readable side.

8.9. A transform stream that replaces template tags

It’s often useful to substitute tags with variables on a stream of data, where the parts that need to be replaced are small compared to the overall data size. This example presents a simple way to do that. It maps strings to strings, transforming a template like "Time: {{time}} Message: {{message}}" to "Time: 15:36 Message: hello" assuming that { time: "15:36", message: "hello" } was passed in the substitutions parameter to LipFuzzTransformer.

This example also demonstrates one way to deal with a situation where a chunk contains partial data that cannot be transformed until more data is received. In this case, a partial template tag will be accumulated in the partialChunk instance variable until either the end of the tag is found or the end of the stream is reached.

class LipFuzzTransformer {
  constructor(substitutions) {
    this.substitutions = substitutions;
    this.partialChunk = "";
    this.lastIndex = undefined;
  }

  transform(chunk, controller) {
    chunk = this.partialChunk + chunk;
    this.partialChunk = "";
    // lastIndex is the index of the first character after the last substitution.
    this.lastIndex = 0;
    chunk = chunk.replace(/\{\{([a-zA-Z0-9_-]+)\}\}/g, this.replaceTag.bind(this));
    // Regular expression for an incomplete template at the end of a string.
    const partialAtEndRegexp = /\{(\{([a-zA-Z0-9_-]+(\})?)?)?$/g;
    // Avoid looking at any characters that have already been substituted.
    partialAtEndRegexp.lastIndex = this.lastIndex;
    this.lastIndex = undefined;
    const match = partialAtEndRegexp.exec(chunk);
    if (match) {
      this.partialChunk = chunk.substring(match.index);
      chunk = chunk.substring(0, match.index);
    }
    controller.enqueue(chunk);
  }

  flush(controller) {
    if (this.partialChunk.length > 0) {
      controller.enqueue(this.partialChunk);
    }
  }

  replaceTag(match, p1, offset) {
    let replacement = this.substitutions[p1];
    if (replacement === undefined) {
      replacement = "";
    }
    this.lastIndex = offset + replacement.length;
    return replacement;
  }
}

In this case we define the transformer to be passed to the TransformStream constructor as a class. This is useful when there is instance data to track.

The class would be used in code like:

const data = { userName, displayName, icon, date };
const ts = new TransformStream(new LipFuzzTransformer(data));

fetchEvent.respondWith(
  fetch(url).then(response => {
    const transformedBody = response.body.pipeThrough(ts);
    return new Response(transformedBody);
  )
);
For simplicity, LipFuzzTransformer performs unescaped text substitutions. In real applications, a template system that performs context-aware escaping is good practice for security and robustness.

8.10. A transform stream created from a sync mapper function

The following function allows creating new TransformStream instances from synchronous "mapper" functions, of the type you would normally pass to Array.prototype.map. It demonstrates that the API is concise even for trivial transforms.

function mapperTransformStream(mapperFunction) {
  return new TransformStream({
    transform(chunk, controller) {
      controller.enqueue(mapperFunction(chunk));
    }
  });
}

This function can then be used to create a TransformStream that uppercases all its inputs:

const ts = mapperTransformStream(chunk => chunk.toUpperCase());
const writer = ts.writable.getWriter();
const reader = ts.readable.getReader();

writer.write("No need to shout");

// Logs "NO NEED TO SHOUT":
reader.read().then(({ value }) => console.log(value));

Although a synchronous transform never causes backpressure itself, it will only transform chunks as long as there is no backpressure, so resources will not be wasted.

Exceptions error the stream in a natural way:

const ts = mapperTransformStream(chunk => JSON.parse(chunk));
const writer = ts.writable.getWriter();
const reader = ts.readable.getReader();

writer.write("[1, ");

// Logs a SyntaxError, twice:
reader.read().catch(e => console.error(e));
writer.write("{}").catch(e => console.error(e));

Conventions

This specification uses algorithm conventions very similar to those of [ECMASCRIPT], whose rules should be used to interpret it (apart from the exceptions enumerated below). In particular, the objects specified here should be treated as built-in objects. For example, their name and length properties are derived as described by that specification, as are the default property descriptor values and the treatment of missing, undefined, or surplus arguments.

We also depart from the [ECMASCRIPT] conventions in the following ways, mostly for brevity. It is hoped (and vaguely planned) that the conventions of ECMAScript itself will evolve in these ways.

It’s also worth noting that, as in [ECMASCRIPT], all numbers are represented as double-precision floating point values, and all arithmetic operations performed on them must be done in the standard way for such values.

Acknowledgments

The editors would like to thank Anne van Kesteren, Ben Kelly, Bert Belder, Brian di Palma, Calvin Metcalf, Dominic Tarr, Ed Hager, Forbes Lindesay, Forrest Norvell, Gorgi Kosev, 贺师俊 (hax), Isaac Schlueter, isonmad, Jake Archibald, Jake Verbaten, Janessa Det, Jens Nockert, Mangala Sadhu Sangeet Singh Khalsa, Marcos Caceres, Marvin Hagemeister, Michael Mior, Mihai Potra, Romain Bellessort, Simon Menke, Stephen Sugden, Tab Atkins, Tanguy Krotoff, Thorsten Lorenz, Till Schneidereit, Tim Caswell, Trevor Norris, tzik, Will Chan, Youenn Fablet, 平野裕 (Yutaka Hirano), and Xabier Rodríguez for their contributions to this specification. Community involvement in this specification has been above and beyond; we couldn’t have done it without you.

This standard is written by Adam Rice (Google, ricea@chromium.org), Domenic Denicola (Google, d@domenic.me), and 吉野剛史 (Takeshi Yoshino, Google, tyoshino@chromium.org).

Per CC0, to the extent possible under law, the editors have waived all copyright and related or neighboring rights to this work.

Index

Terms defined by this specification

Terms defined by reference

References

Normative References

[ECMASCRIPT]
ECMAScript Language Specification. URL: https://tc39.github.io/ecma262/
[HTML]
Anne van Kesteren; et al. HTML Standard. Living Standard. URL: https://html.spec.whatwg.org/multipage/
[PROMISES-GUIDE]
Domenic Denicola. Writing Promise-Using Specifications. 16 February 2016. Finding of the W3C TAG. URL: https://www.w3.org/2001/tag/doc/promises-guide

Informative References

[FETCH]
Anne van Kesteren. Fetch Standard. Living Standard. URL: https://fetch.spec.whatwg.org/
[SERVICE-WORKERS]
Alex Russell; et al. Service Workers 1. URL: https://w3c.github.io/ServiceWorker/v1/