feat: tag S3 client user agent for request attribution

[goanpeca]

Reason for change

Label Studio talks to object storage through boto3 in label_studio/io_storages/s3/utils.py. When it builds the S3 client and resource, it does not set a user_agent_extra, so every request goes out with only the default botocore user agent.

This makes it hard to identify Label Studio traffic against Amazon S3 or any S3-compatible object store (for example Backblaze B2, Cloudflare R2, or MinIO, all reached through the same code path via a custom endpoint_url). A stable client marker helps storage-side observability and support when diagnosing connection or throughput issues, without changing any request behavior.

The change attaches a label-studio/<version> token to user_agent_extra on the S3 client and resource. It follows the common framework/version user-agent convention, resolves the version from the installed package metadata, and falls back to label-studio/dev when the package is not installed (source checkout).

Screenshots

N/A. Backend-only change, no UI impact.

Rollout strategy

No feature flag required. The change is purely additive: it appends a token to the S3 client and resource user agent and leaves signature_version (s3v4) and the optional custom endpoint_url untouched. The client and resource now share a single boto3.session.Config instead of constructing two identical ones. No new configuration, environment variable, or dependency is introduced (importlib.metadata is standard library).

Testing

Added a unit test module label_studio/io_storages/s3/tests/test_user_agent.py that verifies:

• _get_user_agent_extra() returns a label-studio/ prefixed token.
• both the client and the resource carry that token in user_agent_extra.
• signature_version stays s3v4.
• a custom endpoint_url is preserved end to end (using an S3-compatible endpoint as one example, alongside the default AWS S3 path).

Acceptance criteria: when Label Studio issues an S3 request (against AWS S3 or an S3-compatible endpoint), the outgoing user agent contains a label-studio/<version> token, and signature version and endpoint configuration are unchanged.

Run locally:

cd label_studio
pytest io_storages/s3/tests/test_user_agent.py

Risks

Low. The token only extends the user-agent string that boto3 already sends; it does not alter authentication, signing, or the request path. The existing test_resolve_s3_url.py behavior is unaffected. If package metadata cannot be resolved, the code degrades to label-studio/dev rather than raising.

Reviewer notes

Scope is intentionally minimal and additive (two files, one small helper plus the shared Config), in line with the contributing guide's preference for small, single-purpose PRs. The PR title uses the feat: prefix per the repository's title convention. Happy to guard this behind a flag or adjust the token format if you would prefer a different convention.

[netlify]

👷 Deploy request for label-studio-docs-new-theme pending review.

Visit the deploys page to approve it

Name	Link
🔨 Latest commit	0b3846f

[netlify]

👷 Deploy request for heartex-docs pending review.

Visit the deploys page to approve it

Name	Link
🔨 Latest commit	0b3846f

[netlify]

✅ Deploy Preview for label-studio-storybook canceled.

Name	Link
🔨 Latest commit	0b3846f
🔍 Latest deploy log	https://app.netlify.com/projects/label-studio-storybook/deploys/6a451d5a254e230007fa991e

[netlify]

✅ Deploy Preview for label-studio-playground canceled.

Name	Link
🔨 Latest commit	0b3846f
🔍 Latest deploy log	https://app.netlify.com/projects/label-studio-playground/deploys/6a451d5a27235f0008f3fc4b