Skip to content

feat(util-stream): buffering of stream chunks #1523

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Feb 14, 2025
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 5 additions & 0 deletions .changeset/thirty-walls-pump.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,5 @@
---
"@smithy/util-stream": minor
---

utility for buffering stream chunks
5 changes: 4 additions & 1 deletion packages/util-stream/package.json
Original file line number Diff line number Diff line change
Expand Up @@ -61,6 +61,7 @@
"browser": {
"./dist-es/checksum/ChecksumStream": "./dist-es/checksum/ChecksumStream.browser",
"./dist-es/checksum/createChecksumStream": "./dist-es/checksum/createChecksumStream.browser",
"./dist-es/createBufferedReadable": "./dist-es/createBufferedReadableStream",
"./dist-es/getAwsChunkedEncodingStream": "./dist-es/getAwsChunkedEncodingStream.browser",
"./dist-es/headStream": "./dist-es/headStream.browser",
"./dist-es/sdk-stream-mixin": "./dist-es/sdk-stream-mixin.browser",
Expand All @@ -73,12 +74,14 @@
"./dist-es/sdk-stream-mixin": "./dist-es/sdk-stream-mixin.browser",
"./dist-es/headStream": "./dist-es/headStream.browser",
"./dist-es/splitStream": "./dist-es/splitStream.browser",
"./dist-es/createBufferedReadable": "./dist-es/createBufferedReadableStream",
"./dist-cjs/checksum/createChecksumStream": "./dist-cjs/checksum/createChecksumStream.browser",
"./dist-cjs/checksum/ChecksumStream": "./dist-cjs/checksum/ChecksumStream.browser",
"./dist-cjs/getAwsChunkedEncodingStream": "./dist-cjs/getAwsChunkedEncodingStream.browser",
"./dist-cjs/sdk-stream-mixin": "./dist-cjs/sdk-stream-mixin.browser",
"./dist-cjs/headStream": "./dist-cjs/headStream.browser",
"./dist-cjs/splitStream": "./dist-cjs/splitStream.browser"
"./dist-cjs/splitStream": "./dist-cjs/splitStream.browser",
"./dist-cjs/createBufferedReadable": "./dist-cjs/createBufferedReadableStream"
},
"homepage": "https://github.com/awslabs/smithy-typescript/tree/main/packages/util-stream",
"repository": {
Expand Down
37 changes: 37 additions & 0 deletions packages/util-stream/src/ByteArrayCollector.ts
Original file line number Diff line number Diff line change
@@ -0,0 +1,37 @@
/**
* Aggregates byteArrays on demand.
* @internal
*/
export class ByteArrayCollector {
public byteLength = 0;
private byteArrays = [] as Uint8Array[];

public constructor(public readonly allocByteArray: (size: number) => Uint8Array) {}

public push(byteArray: Uint8Array) {
this.byteArrays.push(byteArray);
this.byteLength += byteArray.byteLength;
}

public flush() {
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

will flush() always create a new aggregate buffer, even if there's only one chunk?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

updated to return single bytearray without copy

if (this.byteArrays.length === 1) {
const bytes = this.byteArrays[0];
this.reset();
return bytes;
}
const aggregation = this.allocByteArray(this.byteLength);
let cursor = 0;
for (let i = 0; i < this.byteArrays.length; ++i) {
const bytes = this.byteArrays[i];
aggregation.set(bytes, cursor);
cursor += bytes.byteLength;
}
this.reset();
return aggregation;
}

private reset() {
this.byteArrays = [];
this.byteLength = 0;
}
}
76 changes: 76 additions & 0 deletions packages/util-stream/src/createBufferedReadable.spec.ts
Original file line number Diff line number Diff line change
@@ -0,0 +1,76 @@
import { Readable } from "node:stream";
import { describe, expect, test as it, vi } from "vitest";

import { createBufferedReadable } from "./createBufferedReadable";
import { headStream } from "./headStream";

describe("Buffered Readable stream", () => {
function stringStream(size: number, chunkSize: number) {
async function* generate() {
while (size > 0) {
yield "a".repeat(chunkSize);
size -= chunkSize;
}
}
return Readable.from(generate());
}
function byteStream(size: number, chunkSize: number) {
async function* generate() {
while (size > 0) {
yield Buffer.from(new Uint8Array(chunkSize));
size -= chunkSize;
}
}
return Readable.from(generate());
}
const logger = {
debug: vi.fn(),
info: vi.fn(),
warn: vi.fn(),
error() {},
};

it("should join upstream chunks if they are too small (stringStream)", async () => {
const upstream = stringStream(1024, 8);
const downstream = createBufferedReadable(upstream, 64);

let upstreamChunkCount = 0;
upstream.on("data", () => {
upstreamChunkCount += 1;
});

let downstreamChunkCount = 0;
downstream.on("data", () => {
downstreamChunkCount += 1;
});

await headStream(downstream, Infinity);

expect(upstreamChunkCount).toEqual(128);
expect(downstreamChunkCount).toEqual(16);
});

it("should join upstream chunks if they are too small (byteStream)", async () => {
const upstream = byteStream(1031, 7);
const downstream = createBufferedReadable(upstream, 49, logger);

let upstreamChunkCount = 0;
upstream.on("data", () => {
upstreamChunkCount += 1;
});

let downstreamChunkCount = 0;
downstream.on("data", () => {
downstreamChunkCount += 1;
});

await headStream(downstream, Infinity);

expect(Math.ceil(1031 / 7)).toBe(148);
expect(Math.ceil(1031 / 49)).toBe(22);

expect(upstreamChunkCount).toEqual(148);
expect(downstreamChunkCount).toEqual(22);
expect(logger.warn).toHaveBeenCalled();
});
});
78 changes: 78 additions & 0 deletions packages/util-stream/src/createBufferedReadable.ts
Original file line number Diff line number Diff line change
@@ -0,0 +1,78 @@
import type { Logger } from "@smithy/types";
import { Readable } from "node:stream";

import { ByteArrayCollector } from "./ByteArrayCollector";
import type { BufferStore, Modes } from "./createBufferedReadableStream";
import { createBufferedReadableStream, flush, merge, modeOf, sizeOf } from "./createBufferedReadableStream";
import { isReadableStream } from "./stream-type-check";

/**
* @internal
* @param upstream - any Readable or ReadableStream.
* @param size - byte or character length minimum. Buffering occurs when a chunk fails to meet this value.
* @param onBuffer - for emitting warnings when buffering occurs.
* @returns another stream of the same data and stream class, but buffers chunks until
* the minimum size is met, except for the last chunk.
*/
export function createBufferedReadable(upstream: Readable, size: number, logger?: Logger): Readable;
export function createBufferedReadable(upstream: ReadableStream, size: number, logger?: Logger): ReadableStream;
export function createBufferedReadable(
upstream: Readable | ReadableStream,
size: number,
logger?: Logger
): Readable | ReadableStream {
if (isReadableStream(upstream)) {
return createBufferedReadableStream(upstream, size, logger);
}
const downstream = new Readable({ read() {} });
let streamBufferingLoggedWarning = false;
let bytesSeen = 0;

const buffers = [
"",
new ByteArrayCollector((size) => new Uint8Array(size)),
new ByteArrayCollector((size) => Buffer.from(new Uint8Array(size))),
] as BufferStore;
let mode: Modes | -1 = -1;

upstream.on("data", (chunk) => {
const chunkMode = modeOf(chunk);
if (mode !== chunkMode) {
if (mode >= 0) {
downstream.push(flush(buffers, mode));
}
mode = chunkMode;
}
if (mode === -1) {
downstream.push(chunk);
return;
}

const chunkSize = sizeOf(chunk);
bytesSeen += chunkSize;
const bufferSize = sizeOf(buffers[mode]);
if (chunkSize >= size && bufferSize === 0) {
// skip writing to the intermediate buffer
// because the upstream chunk is already large enough.
downstream.push(chunk);
} else {
// buffer and potentially flush the data downstream.
const newSize = merge(buffers, mode, chunk);
if (!streamBufferingLoggedWarning && bytesSeen > size * 2) {
streamBufferingLoggedWarning = true;
logger?.warn(
`@smithy/util-stream - stream chunk size ${chunkSize} is below threshold of ${size}, automatically buffering.`
);
}
if (newSize >= size) {
downstream.push(flush(buffers, mode));
}
}
});
upstream.on("end", () => {
downstream.push(flush(buffers, mode));
downstream.push(null);
});

return downstream;
}
Original file line number Diff line number Diff line change
@@ -0,0 +1,107 @@
import { Readable } from "node:stream";
import { describe, expect, test as it, vi } from "vitest";

import { createBufferedReadable } from "./createBufferedReadableStream";

describe("Buffered ReadableStream", () => {
function stringStream(size: number, chunkSize: number) {
async function* generate() {
while (size > 0) {
yield "a".repeat(chunkSize);
size -= chunkSize;
}
}
return Readable.toWeb(Readable.from(generate()));
}
function byteStream(size: number, chunkSize: number) {
async function* generate() {
while (size > 0) {
yield new Uint8Array(chunkSize);
size -= chunkSize;
}
}
return Readable.toWeb(Readable.from(generate()));
}

const logger = {
debug: vi.fn(),
info: vi.fn(),
warn: vi.fn(),
error() {},
};

it("should join upstream chunks if they are too small (stringStream)", async () => {
let upstreamChunkCount = 0;
let downstreamChunkCount = 0;

const upstream = stringStream(1024, 8);
const upstreamReader = upstream.getReader();

const midstream = new ReadableStream({
async pull(controller) {
const { value, done } = await upstreamReader.read();
if (done) {
controller.close();
} else {
expect(value.length).toBe(8);
upstreamChunkCount += 1;
controller.enqueue(value);
}
},
});
const downstream = createBufferedReadable(midstream, 64);
const reader = downstream.getReader();

while (true) {
const { done, value } = await reader.read();
if (done) {
break;
} else {
downstreamChunkCount += 1;
expect(value.length).toBe(64);
}
}

expect(upstreamChunkCount).toEqual(128);
expect(downstreamChunkCount).toEqual(16);
});

it("should join upstream chunks if they are too small (byteStream)", async () => {
let upstreamChunkCount = 0;
let downstreamChunkCount = 0;

const upstream = byteStream(1031, 7);
const upstreamReader = upstream.getReader();

const midstream = new ReadableStream({
async pull(controller) {
const { value, done } = await upstreamReader.read();
if (done) {
controller.close();
} else {
expect(value.length).toBe(7);
upstreamChunkCount += 1;
controller.enqueue(value);
}
},
});
const downstream = createBufferedReadable(midstream, 49, logger);
const downstreamReader = downstream.getReader();

while (true) {
const { done, value } = await downstreamReader.read();
if (done) {
break;
} else {
downstreamChunkCount += 1;
if (value.byteLength > 7) {
expect(value.byteLength).toBe(49);
}
}
}

expect(upstreamChunkCount).toEqual(148);
expect(downstreamChunkCount).toEqual(22);
expect(logger.warn).toHaveBeenCalled();
});
});
Loading