fix(server): preserve JSON rejection statuses

[ting-hong-shieh]

What

• preserve the status code carried by Axum JsonRejection
• keep non-object and protocol-decode body errors at HTTP 400
• retain the existing OpenAI-compatible and Anthropic error renderers
• cover malformed JSON, missing/unsupported content type, and oversized bodies

Why

The shared JSON helper reduced every extractor rejection to a string, and invalid_body_error then hard-coded HTTP 400. Clients therefore lost Axum's 415 and 413 distinctions, and Anthropic's existing request_too_large mapping was unreachable for oversized JSON bodies.

Closes #400

Validation

• cargo test -p switchyard-server --test server json_extractor_statuses_keep_api_specific_error_envelopes -- --nocapture — passed
• cargo test -p switchyard-server — passed
• cargo fmt --all --check — passed
• cargo clippy --workspace --all-targets -- -D warnings — passed
• cargo test --workspace — passed with loopback access enabled for the repository's local mock servers
• git diff --check — passed

No live provider calls or secrets were used.

Summary by CodeRabbit

• Bug Fixes
  • Invalid request bodies now preserve the appropriate HTTP status codes instead of being returned as generic bad requests.
  • Improved error responses for malformed JSON, unsupported media types, and oversized request bodies.
  • Consistent API-specific error envelopes are now returned across Chat Completions and Anthropic Messages endpoints.

[ting-hong-shieh]

Before / after HTTP snapshot

I ran the same focused in-process Axum request matrix at the PR base (b256d936) and current head (caed4eff). The requests are rejected by extraction before routing, so no upstream/provider call was involved.

OpenAI Chat endpoint:

Input	Before status	After status	After envelope
malformed JSON + application/json	400	400	invalid_request_error / invalid_body
valid JSON, missing Content-Type	400	415	invalid_request_error / invalid_body
valid JSON + unsupported Content-Type	400	415	invalid_request_error / invalid_body
body larger than 32 MiB	400	413	invalid_request_error / invalid_body

Anthropic Messages endpoint:

Input	Before status/type	After status/type
malformed JSON + application/json	400 / invalid_request_error	400 / invalid_request_error
valid JSON, missing Content-Type	400 / invalid_request_error	415 / api_error
body larger than 32 MiB	400 / invalid_request_error	413 / request_too_large

The outer Anthropic envelope remains {"type":"error", ...}; the change restores Axum's rejection status and lets the existing protocol renderer select the status-appropriate error type.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 7b07d603-bcc0-4e64-aff8-73758531b2da

📥 Commits

Reviewing files that changed from the base of the PR and between b256d93 and caed4ef.

📒 Files selected for processing (2)

• crates/switchyard-server/src/lib.rs
• crates/switchyard-server/tests/server.rs

---

Walkthrough

The server now preserves JSON extractor HTTP statuses, including 415 Unsupported Media Type and 413 Payload Too Large. Tests cover malformed JSON, unsupported media types, oversized bodies, and API-specific error envelopes.

Changes

Request body error handling

Layer / File(s)	Summary
Preserve extractor status codes crates/switchyard-server/src/lib.rs	Body decoding now returns the extractor status with its error message. Invalid-body responses use the supplied status, while non-object JSON and route decoding errors remain 400 Bad Request.
Validate request body responses crates/switchyard-server/tests/server.rs	Added raw-body request support and tests for malformed JSON, missing or unsupported content types, and oversized bodies across Chat Completions and Anthropic Messages endpoints.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Mergeability Score: ⚪ Minimal · up to caed4

The change preserves API-specific JSON rejection statuses while retaining existing error envelopes, with no actionable merge-blocking risk remaining after normal checks and review.

Poem

I’m a rabbit guarding JSON at night,
Keeping four-oh-four? No—statuses right.
Four-one-five for types astray,
Four-one-three when bytes outweigh.
Tests hop through each error door,
And envelopes say what they’re for. 🐇

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and concisely describes preserving JSON rejection HTTP statuses, which is the primary change.
Linked Issues check	✅ Passed	The changes satisfy issue #400 by preserving 415, 413, and 400 statuses while retaining OpenAI and Anthropic error mappings.
Out of Scope Changes check	✅ Passed	The implementation and regression tests directly support issue #400 and contain no unrelated changes.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

---

_{Comment @coderabbitai help to get the list of available commands.}