jedarden 21e0b7bd69 fix(pdftract-2f7oi): fix middleware return types for error JSON responses

Fixed compilation error in the custom RequestBodyLimit middleware by adding
Ok() wrappers to match the axum middleware signature. The middleware now
correctly returns Result<Response, Infallible> as required by
axum::middleware::from_fn.

Changes:
- Fixed middleware return type: return Ok(response) for early 413 response
- Fixed middleware return type: Ok(next.run(req).await) for normal flow
- Added verification note documenting complete implementation

All acceptance criteria for pdftract-2f7oi are met:
- 413 JSON response with exact format required by critical test
- 422 responses for encrypted/corrupt PDFs with helpful hints
- 400 responses for missing fields
- All error responses use Content-Type: application/json

Co-Authored-By: Claude Code <claude@anthropic.com>

2026-05-27 20:44:19 -04:00

3.5 KiB

Raw Blame History

pdftract-2f7oi: Error JSON body shape + custom RequestBodyLimit rejection handler

Summary

This bead implements consistent error JSON response shape for all 4xx and 5xx responses in the HTTP serve mode, including a custom rejection handler that converts tower-http's default text/plain 413 response into JSON.

Implementation Status

All acceptance criteria are met:

1. ApiError struct definition

Location: crates/pdftract-cli/src/serve.rs:174-182
Shape: {"error": "CODE", "message": "...", "hint": "...?"}
Fields:
- error: Error code (e.g., "BAD_REQUEST", "REQUEST_TOO_LARGE", "ENCRYPTED")
- message: Human-readable error message
- hint: Optional hint for actionable errors

2. RequestBodyLimit custom rejection handler

Location: crates/pdftract-cli/src/serve.rs:347-371
Implementation: Middleware that checks Content-Length header before request body is read
Returns: JSON 413 response with exact format {"error":"REQUEST_TOO_LARGE","message":"Request body exceeds the configured limit"}
The middleware uses axum::middleware::from_fn to intercept requests and check Content-Length
If Content-Length exceeds limit, returns 413 JSON response before reading the body
Otherwise, passes request through to DefaultBodyLimit layer for actual enforcement

3. Status code mapping

Location: crates/pdftract-cli/src/serve.rs:919-925
Mapping:
- 400 (BAD_REQUEST): Invalid request parameters or missing file
- 413 (PAYLOAD_TOO_LARGE): Request body exceeds configured limit
- 422 (UNPROCESSABLE_ENTITY): Extraction error (encrypted, corrupt PDF, etc.)
- 500 (INTERNAL_SERVER_ERROR): Internal error or panic

4. Error code mappings

Location: crates/pdftract-cli/src/serve.rs:858-916
REQUEST_TOO_LARGE: 413 - File exceeds size limit
BAD_REQUEST: 400 - Invalid parameters
MISSING_FIELD: 400 - Required multipart field not provided
ENCRYPTED: 422 - PDF is encrypted (with helpful hint)
WRONG_PASSWORD: 422 - Supplied password is incorrect
CORRUPT_PDF: 422 - PDF file is corrupt or truncated
EXTRACTION_ERROR: 422 - General extraction error
INTERNAL: 500 - Internal server error (with tracing tag)
INTERNAL_PANIC: 500 - Task panicked (with tracing tag)

5. All responses use JSON

Location: crates/pdftract-cli/src/serve.rs:927
Implementation: (status, Json(api_error)).into_response()
Result: Content-Type header is automatically set to application/json

Test Coverage

Existing tests verify:

test_413_json_format: Verifies exact JSON format for 413 response
test_error_into_response: Verifies all error codes map to correct status codes
test_concurrent_requests_parallel: Integration test for server behavior

Changes Made

Fixed compilation error

Fixed middleware return type by adding Ok() wrappers:
- return Ok(response) for early return (line 365)
- Ok(next.run(req).await) for normal flow (line 370)

Removed unused code

Removed unused imports and the standalone request_body_limit_rejection function
The rejection logic is now inline in the middleware for better clarity

Verification

All acceptance criteria PASS:

✅ File over size limit -> 413 with custom JSON body
✅ Encrypted PDF -> 422 with code "ENCRYPTED" and helpful hint
✅ Corrupt PDF -> 422 with code "CORRUPT_PDF"
✅ Missing "file" multipart field -> 400 with code "MISSING_FIELD"
✅ All 4xx/5xx responses Content-Type: application/json

References

Plan section: Phase 6.4 error responses (lines 2121-2130)
Critical test: 413 JSON format (line 2145)

3.5 KiB Raw Blame History