TransformStream: TransformStream() constructor

transformer Optional

An object representing the transformer. If not supplied the resulting stream will be an identity transform stream which forwards all chunks written to its writable side to its readable side, without any changes.

The transformer object can contain any of the following methods. In each method controller is an instance of TransformStreamDefaultController.

start(controller)

Called when the TransformStream is constructed. It is typically used to enqueue chunks using TransformStreamDefaultController.enqueue().

transform(chunk, controller)

Called when a chunk written to the writable side is ready to be transformed, and performs the work of the transformation stream. It can return a promise to signal success or failure of the write operation. If no transform() method is supplied, the identity transform is used, and the chunk will be enqueued with no changes.

flush(controller)

Called after all chunks written to the writable side have been successfully transformed, and the writable side is about to be closed.

writableStrategy Optional

An object that optionally defines a queuing strategy for the stream. This takes two parameters:

highWaterMark

A non-negative integer. This defines the total number of chunks that can be contained in the internal queue before backpressure is applied.

size(chunk)

A method containing a parameter chunk. This indicates the size to use for each chunk, in bytes.

readableStrategy Optional

An object that optionally defines a queuing strategy for the stream. This takes two parameters:

highWaterMark

A non-negative integer. This defines the total number of chunks that can be contained in the internal queue before backpressure is applied.

size(chunk)

A method containing a parameter chunk. This indicates the size to use for each chunk, in bytes.

The following example transforms text to uppercase chunk by chunk. This example is from Streams—The Definitive Guide, which has a number of examples of different types of streams.

function upperCaseStream() {
  return new TransformStream({
    transform(chunk, controller) {
      controller.enqueue(chunk.toUpperCase());
    },
  });
}

function appendToDOMStream(el) {
  return new WritableStream({
    write(chunk) {
      el.append(chunk);
    },
  });
}

fetch("./lorem-ipsum.txt").then((response) =>
  response.body
    .pipeThrough(new TextDecoderStream())
    .pipeThrough(upperCaseStream())
    .pipeTo(appendToDOMStream(document.body)),
);

If no transformer argument is supplied then the result will be an identity transform stream which forwards all chunks written to the writable side to the readable side with no changes. In the following example an identity transform stream is used to add buffering to a pipe.

const writableStrategy = new ByteLengthQueuingStrategy({
  highWaterMark: 1024 * 1024,
});
readableStream
  .pipeThrough(new TransformStream(undefined, writableStrategy))
  .pipeTo(writableStream);