Merged
Conversation
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (1)
🚧 Files skipped from review as they are similar to previous changes (1)
📝 WalkthroughWalkthroughAdds a new ResponsesApiFormat ChatFormat that bridges OpenAI Responses API requests/responses to the internal chat completions hub, implements streaming/tool-call accumulation and native delegation, and updates native streaming state to carry usage. ChangesResponses API Format Bridge
Sequence DiagramsequenceDiagram
participant Client as Responses API<br/>Client
participant Format as ResponsesApiFormat
participant Hub as Chat Completions<br/>Hub
participant Native as Native OpenAI<br/>Handlers
Client->>Format: ResponsesApiRequest (instructions, items, tools)
Format->>Format: validate_request()
Format->>Format: to_hub() -> ChatCompletionRequest
alt Native Delegation
Format->>Native: call_native(request)
Native->>Native: transform_request()
Native->>Hub: send native request
Hub-->>Native: ChatCompletionResponse / stream
Native->>Native: transform_response()
Native-->>Format: transformed response / stream events
else Direct Hub Call
Format->>Hub: send ChatCompletionRequest
Hub-->>Format: ChatCompletionResponse / stream
end
Format->>Format: from_hub() -> ResponsesApiResponse
Format-->>Client: ResponsesApiResponse (output_text, function_calls, usage)
alt Streaming
Hub->>Format: ChatCompletionChunk stream
Format->>Format: merge_streaming_tool_call() / accumulate text deltas
Format-->>Client: ResponsesApiStreamEvent (output_item_added / content_part_added / content_block_delta)
Format-->>Client: ResponsesApiStreamEvent (output_item_done / content_block_done)
end
Estimated code review effort🎯 4 (Complex) | ⏱️ ~45 minutes Possibly related PRs
🚥 Pre-merge checks | ✅ 5 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Review rate limit: 3/5 reviews remaining, refill in 23 minutes and 5 seconds. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 2
🧹 Nitpick comments (1)
src/gateway/formats/openai/responses.rs (1)
27-30: ⚡ Quick winAdd
///docs to the new public types.
ResponsesApiFormatis publicly re-exported, andResponsesBridgeStateis public as well, but neither has a doc comment.Minimal doc-comment patch
+/// Bridges OpenAI Responses API requests and streams through the hub chat format. pub struct ResponsesApiFormat; #[derive(Debug, Clone, Default)] +/// Stateful data used while converting hub streaming chunks into Responses API events. pub struct ResponsesBridgeState {As per coding guidelines "Use /// for doc comments on public items in Rust".
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@src/gateway/formats/openai/responses.rs` around lines 27 - 30, Add /// doc comments to the new public types: place a short descriptive triple-slash comment above the ResponsesApiFormat declaration explaining its role as the API response format adapter, and above the ResponsesBridgeState struct explaining what state it holds and when it is used; ensure the comments are concise, use proper Rust doc style (///), describe public behavior or purpose, and mention any important invariants or usage notes for ResponsesApiFormat and ResponsesBridgeState.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@src/gateway/formats/openai/responses.rs`:
- Around line 204-258: The stream assigns different numeric indices to the same
logical output as items (tools/text) are added, because tool_output_index(state,
...) recomputes index from current presence; fix by making the output index
stable and stored in the per-item state: when handling a tool call (functions:
merge_streaming_tool_call, streaming_function_call_output_item, and where you
currently call tool_output_index), check the tool_state for a stored
output_index Option, and if None compute the index once (using the same
logic/tool_output_index helper) and persist it to tool_state.output_index before
emitting any
OutputItemAdded/OutputTextDelta/FunctionCallArgumentsDelta/Output... events;
replace subsequent calls to tool_output_index with reading
tool_state.output_index so all later delta/done/completed events use the same
stable index. Apply the same pattern in the other referenced blocks (around
lines ~292-305, ~749-786, ~788-823).
- Around line 71-98: The bridge currently accepts requests with
previous_response_id but to_hub() only forwards the current input, so follow-ups
with FunctionCallOutput become orphaned tool messages; fix this by updating
ensure_request_is_bridgeable() to reject any request that has
previous_response_id (and/or any ResponsesInput item that represents a
FunctionCallOutput continuation) by returning an Err when
req.previous_response_id.is_some() (and the analogous predicate for input
items), so to_hub(), responses_input_item_to_hub_message, and BridgeContext are
never invoked for these unresolved continuations; reference
ensure_request_is_bridgeable, to_hub, BridgeContext, previous_response_id, and
responses_input_item_to_hub_message when making the change.
---
Nitpick comments:
In `@src/gateway/formats/openai/responses.rs`:
- Around line 27-30: Add /// doc comments to the new public types: place a short
descriptive triple-slash comment above the ResponsesApiFormat declaration
explaining its role as the API response format adapter, and above the
ResponsesBridgeState struct explaining what state it holds and when it is used;
ensure the comments are concise, use proper Rust doc style (///), describe
public behavior or purpose, and mention any important invariants or usage notes
for ResponsesApiFormat and ResponsesBridgeState.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro
Run ID: 157487af-ca4d-4937-bd88-a5868193fa8e
📒 Files selected for processing (4)
src/gateway/formats/mod.rssrc/gateway/formats/openai/mod.rssrc/gateway/formats/openai/responses.rssrc/gateway/traits/native.rs
Comment on lines
+71
to
+98
| fn to_hub(req: &Self::Request) -> Result<(ChatCompletionRequest, BridgeContext)> { | ||
| ensure_request_is_bridgeable(req)?; | ||
|
|
||
| let mut messages = Vec::new(); | ||
| if let Some(instructions) = req.instructions.as_ref().filter(|text| !text.is_empty()) { | ||
| messages.push(ChatMessage { | ||
| role: "system".into(), | ||
| content: Some(MessageContent::Text(instructions.clone())), | ||
| name: None, | ||
| tool_calls: None, | ||
| tool_call_id: None, | ||
| }); | ||
| } | ||
|
|
||
| match &req.input { | ||
| ResponsesInput::Text(text) => messages.push(ChatMessage { | ||
| role: "user".into(), | ||
| content: Some(MessageContent::Text(text.clone())), | ||
| name: None, | ||
| tool_calls: None, | ||
| tool_call_id: None, | ||
| }), | ||
| ResponsesInput::Items(items) => { | ||
| for item in items { | ||
| messages.push(responses_input_item_to_hub_message(item)?); | ||
| } | ||
| } | ||
| } |
There was a problem hiding this comment.
Reject previous_response_id tool-result continuations until the bridge can reconstruct history.
to_hub() only forwards the current input, while previous_response_id is just echoed into BridgeContext. A follow-up request containing FunctionCallOutput therefore turns into a standalone role="tool" message with no preceding assistant tool_calls message in the Chat Completions payload, which will break multi-turn tool-calling flows.
Either resolve previous_response_id into prior messages before building the hub request, or mark these requests as non-bridgeable in ensure_request_is_bridgeable() for now.
Temporary guard if history reconstruction is not ready yet
fn ensure_request_is_bridgeable(request: &ResponsesApiRequest) -> Result<()> {
+ let has_function_call_output = matches!(
+ &request.input,
+ ResponsesInput::Items(items)
+ if items.iter().any(|item| matches!(item, ResponsesInputItem::FunctionCallOutput { .. }))
+ );
+
+ if request.previous_response_id.is_some() || has_function_call_output {
+ return Err(GatewayError::Bridge(
+ "Responses API continuation/tool-result turns cannot be bridged through Chat Completions yet"
+ .into(),
+ ));
+ }
+
if request.background.unwrap_or(false) {
return Err(GatewayError::Bridge(
"Responses API background mode cannot be bridged through Chat Completions".into(),Also applies to: 393-417, 420-437, 564-610
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@src/gateway/formats/openai/responses.rs` around lines 71 - 98, The bridge
currently accepts requests with previous_response_id but to_hub() only forwards
the current input, so follow-ups with FunctionCallOutput become orphaned tool
messages; fix this by updating ensure_request_is_bridgeable() to reject any
request that has previous_response_id (and/or any ResponsesInput item that
represents a FunctionCallOutput continuation) by returning an Err when
req.previous_response_id.is_some() (and the analogous predicate for input
items), so to_hub(), responses_input_item_to_hub_message, and BridgeContext are
never invoked for these unresolved continuations; reference
ensure_request_is_bridgeable, to_hub, BridgeContext, previous_response_id, and
responses_input_item_to_hub_message when making the change.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary by CodeRabbit
New Features
Tests