feat(analytics-browser): support gzip request body compression by Mercy811 · Pull Request #1542 · amplitude/Amplitude-TypeScript

Mercy811 · 2026-02-17T23:06:47Z

Summary

Gzip request if server url is default.
Add a new configuration enableRequestBodyCompression to disable it if server url is set to a proxy url
Same logic as amplitude/Amplitude-Swift@12bbd67
Only payload larger than 2kb will be compressed

Checklist

Does your PR title have the correct title format?
Does your PR have a breaking change?:

Note

Medium Risk
Touches the core event upload pipeline and transport interface; mistakes could break ingestion or proxy setups, though changes are gated by endpoint checks, size thresholds, and extensive tests.

Overview
Adds optional gzip compression for event upload bodies by extending the core Transport.send() API and plumbing a new enableRequestBodyCompression config through browser/core config and the destination upload path.

Compression is forced for Amplitude’s default ingestion endpoints, while custom serverUrls only compress when explicitly enabled; browser transports (XHRTransport and a browser-local FetchTransport) gzip payloads �≥2MB when CompressionStream is available, and SendBeaconTransport explicitly opts out since it can’t set headers. E2E and unit tests are updated/added to handle gzipped bodies (including request parsing helpers) and to validate compression behavior across transports and endpoint types.

^{Written by Cursor Bugbot for commit df2cac1. Configure here.}

packages/analytics-core/src/transports/gzip.ts

test: e2e Co-Authored-By: Cursor <cursoragent@cursor.com>

Mercy811 · 2026-02-20T00:44:18Z

bugbot run

cursor

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

packages/analytics-core/src/transports/fetch.ts

daniel-graham-amplitude · 2026-02-20T01:19:14Z

Because this adds extra computation prior to calling fetch, I think this increases the risk of us missing requests that are made after the pageunload event. Would it be possible to disable compression while the page is unloading?

packages/analytics-browser/src/transports/xhr.ts

packages/analytics-browser/src/config.ts

packages/analytics-core/src/transports/gzip.ts

Mercy811 · 2026-02-26T02:34:38Z

bugbot run

cursor

Cursor Bugbot has reviewed your changes and found 2 potential issues.

^{Bugbot Autofix is OFF. To automatically fix reported issues with Cloud Agents, enable autofix in the Cursor dashboard.}

packages/analytics-browser/src/transports/xhr.ts

packages/analytics-browser/src/transports/fetch.ts

daniel-graham-amplitude · 2026-02-26T05:30:33Z

packages/analytics-browser/src/transports/fetch.ts

+    const shouldCompressBody =
+      shouldCompressUploadBody &&
+      getStringSizeInBytes(bodyString) >= MIN_GZIP_UPLOAD_BODY_SIZE_BYTES &&
+      isCompressionStreamAvailable();


Looking at the getStringSizeInBytes helper functions, it looks like it runs new TextEncoder(value), which creates a copy of the payload and then returns the size, meaning there's an additional copy of the payload in memory (in byte format) that is generated in O(n) time (because it has to iterate through each character in the string and copy it over).

I'm thinking because this threshold is just a heuristic anyways, maybe could just define the minimum upload size as a minimum string length rather than a minimum byte size, then we'd just need to check bodyString.length.

I understand that this goes against our agreement to make it 2kb, but string length should be "good enough". Most characters in a string are 1 byte each, and at most they're 3 bytes each, which gives an upper bound on how much we can under-exaggerate the "size" of a string.

This is a good point! English letter, number, and common punctuation are all 1 byte each. So I'm keeping 2 * 1024 * 1024 and compare it with string length.

daniel-graham-amplitude · 2026-02-26T05:30:56Z

packages/analytics-browser/src/transports/fetch.ts

+      if (compressed) {
+        headers['Content-Encoding'] = 'gzip';
+        body = compressed;
+      }


Awesome! This nice.

Actually I re-thought about this and felt it makes more sense to not allow custom headers override default headers (Content-Type, Accept, Content-Encoding). If it's useless and even harmful to override default headers if transport providers keeps their behavior to set json or gzip payload. Custom headers are support to support proxy authorization in real world for example.

packages/analytics-browser/src/transports/send-beacon.ts

daniel-graham-amplitude · 2026-02-26T05:55:59Z

packages/analytics-browser/src/transports/send-beacon.ts

+    super();
+  }
+
+  async send(serverUrl: string, payload: Payload, _enableRequestBodyCompression = false): Promise<Response | null> {


Since _enableRequestBodyCompression is an optional parameter in the BaseTransport method, you should just be able to leave it out of the send type signature, since it's not being used.

Oh I just wanna make it look aligned in browser platform.

_enableRequestBodyCompression is an optional parameter in the BaseTransport method,

Yes! So we can keep the http transport in node sdk remains unchanged for now

daniel-graham-amplitude · 2026-02-26T06:03:29Z

packages/analytics-browser/src/transports/fetch.ts

+// Temporary browser-specific fetch transport with gzip support.
+// TODO: Merge this implementation back into @amplitude/analytics-core FetchTransport
+// once React Native SDK supports request body gzip.
+export class FetchTransport extends BaseTransport implements Transport {


What was the reason for making a new fetch.ts in analytics-browser? Is it because the core fetch.ts fails to compile in other environments (like React Native), when you add the compression logic?

React Native import core fetch. I don't wanna let it use the gzip feature now because by default we compress

Does React Native fail at runtime or compilation time though? Couldn't we just do a check on !!globalScope.CompressionStream and if it returns false don't run gzip for that platform? I'm just concerned about having 2 separate implementations of fetch that would need to be maintained.

The browser fetch is for short term. After dogfooding, we can have another pr for react native and merge it to the gzip feature branch and then merge gzip feature branch into main together. I tend to avoid making this giant pr bigger.

daniel-graham-amplitude · 2026-02-26T16:52:30Z

packages/analytics-core/src/transports/base.ts


 export class BaseTransport implements Transport {
-  send(_serverUrl: string, _payload: Payload): Promise<Response | null> {
+  send(_serverUrl: string, _payload: Payload, _enableRequestBodyCompression?: boolean): Promise<Response | null> {


Thinking about the signature of this some more. I feel like there's a chance, in the future, that there could be more options that we need to add to send and the method signature would be pretty long.

Would it be much trouble if we did it like this:

type SendOptions = { _enableRequestBodyCompression?: boolean }; send(_serverUrl: string, _payload: Payload, _sendOptions?: SendOptions)

I'll leave this one as "optional".

I thought about this too but realized we probably need a refactor on the whole class. It doesn't make sense to me to set custom headers as a class property but pass server url as a parameter to send. They should be treated the same and as well as the should compress flag.

I’ll keep it as it is because it’s an internal interface, unlike storage provider.

packages/analytics-core/src/types/config/core-config.ts

Mercy811 added 5 commits February 17, 2026 15:06

feat(analytics-browser): support gzip request body compression

e082ac7

test: core

1a065a7

test: browser

7aae69a

fix: duplex

527f866

fix: replace readableStream with arrayBuffer

7ea6c90

daniel-graham-amplitude reviewed Feb 20, 2026

View reviewed changes

packages/analytics-core/src/transports/gzip.ts Outdated Show resolved Hide resolved

Mercy811 force-pushed the AMP-149145-update-browser-sdk-to-use-enable-request-body-compression branch from 3b91b84 to e0f6d6e Compare February 20, 2026 00:28

Revert "test: e2e"

84254a5

test: e2e Co-Authored-By: Cursor <cursoragent@cursor.com>

Mercy811 force-pushed the AMP-149145-update-browser-sdk-to-use-enable-request-body-compression branch from e0f6d6e to 84254a5 Compare February 20, 2026 00:42

cursor bot reviewed Feb 20, 2026

View reviewed changes

daniel-graham-amplitude reviewed Feb 20, 2026

View reviewed changes

packages/analytics-core/src/transports/fetch.ts Outdated Show resolved Hide resolved

daniel-graham-amplitude reviewed Feb 20, 2026

View reviewed changes

packages/analytics-browser/src/transports/xhr.ts Outdated Show resolved Hide resolved

Mercy811 commented Feb 20, 2026

View reviewed changes

packages/analytics-browser/src/config.ts Outdated Show resolved Hide resolved

Mercy811 changed the base branch from main to gzip February 20, 2026 17:53

Mercy811 marked this pull request as ready for review February 20, 2026 18:10

daniel-graham-amplitude reviewed Feb 21, 2026

View reviewed changes

packages/analytics-core/src/transports/gzip.ts Outdated Show resolved Hide resolved

Mercy811 added 6 commits February 25, 2026 15:14

fix(transport): set gzip threshold to 2mb

648d194

calculate shouldCompressUploadBody in destination plugin

a566a33

more

dce5313

browser fetch

19100d4

test: core

e15e09c

test: test

df2cac1

cursor bot reviewed Feb 26, 2026

View reviewed changes

packages/analytics-browser/src/transports/xhr.ts Outdated Show resolved Hide resolved

packages/analytics-browser/src/transports/fetch.ts Show resolved Hide resolved

daniel-graham-amplitude reviewed Feb 26, 2026

View reviewed changes

packages/analytics-browser/src/transports/send-beacon.ts Show resolved Hide resolved

daniel-graham-amplitude reviewed Feb 26, 2026

View reviewed changes

Mercy811 added 2 commits February 25, 2026 23:05

test: more

cb4c4f9

fix: comments

2eb2428

daniel-graham-amplitude reviewed Feb 26, 2026

View reviewed changes

packages/analytics-core/src/types/config/core-config.ts Show resolved Hide resolved

Mercy811 added 2 commits February 26, 2026 10:25

chore: comment

e1b29d4

chore: lint

d1fc50d

Conversation

Mercy811 commented Feb 17, 2026 • edited by daniel-graham-amplitude Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Checklist

Uh oh!

Uh oh!

Mercy811 commented Feb 20, 2026

Uh oh!

cursor bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

daniel-graham-amplitude commented Feb 20, 2026

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Mercy811 commented Feb 26, 2026

Uh oh!

cursor bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

daniel-graham-amplitude Feb 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Mercy811 Feb 26, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Mercy811 commented Feb 17, 2026 •

edited by daniel-graham-amplitude

Loading

daniel-graham-amplitude Feb 26, 2026 •

edited

Loading

Mercy811 Feb 26, 2026 •

edited

Loading