feat(core): expose request body content streams

[sylvesterkaczmarek]

Summary

Add HttpRequestBody.content(): InputStream so custom HTTP client implementations can consume request bodies directly as streams.

Addresses the request-body portion of #664.

Problem

HttpResponse.body() already exposes an InputStream, but HttpRequestBody only exposes writeTo(OutputStream). Custom HttpClient adapters therefore have to bridge the push-style request body API themselves.

Changes

• add a JVM-default content(): InputStream method to HttpRequestBody
• keep existing third-party implementations compatible by providing a default fallback based on writeTo
• override the fallback for built-in JSON and multipart bodies
• stream multipart framing with SequenceInputStream, so InputStream-backed file parts are not buffered into memory
• add focused Kotlin regression coverage for fallback behavior and byte-for-byte multipart parity
• add a Java-facing regression proving existing implementations do not need to implement the new method

Compatibility

The repository compiles Kotlin interfaces with -Xjvm-default=all, so the new method is emitted as a Java default interface method. Existing Java and Kotlin implementations are not required to add an override.

The fallback implementation buffers only for third-party bodies that do not override content(). Built-in SDK request bodies override it, including streaming multipart bodies.

Validation

• branch is based directly on current upstream main at 715a30468d3232a5df4f6c6bec8b2175404eb8b9
• branch is 0 commits behind upstream
• diff is limited to HttpRequestBody, built-in request-body implementations, and focused tests
• focused regression tests are included for Kotlin and Java callers
• full repository validation is left to GitHub Actions because this environment does not have a complete local checkout/toolchain

Addresses #664

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 3d392208bb

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Preserve streaming through the logging body wrapper

When LoggingHttpClient runs at DEBUG level, it replaces the request body with LoggingHttpRequestBody, which does not override content(). A custom HttpClient consuming that body through the new API therefore invokes this fallback, causing LoggingHttpRequestBody.writeTo() to materialize the entire body in a ByteArrayOutputStream before returning; large InputStream-backed multipart uploads consequently lose the streaming behavior introduced here and may exhaust the heap. The logging wrapper needs a streaming content() implementation that preserves its logging behavior.

Useful? React with 👍 / 👎.