The function returns a default ai.ModelResponse if the stream ends without a ResponseCompletedEvent. This can hide potential issues, as a stream should ideally end with either a completion event or an error. Returning a default response might lead to silent failures or confusing behavior for the caller, who would receive an empty response without usage statistics.

Consider returning an error if finalResp is nil and stream.Err() is also nil to make failures more explicit.

@gemini-code-assist This is an unlikely scenario: The user might already have received all the chunks properly, failing at this point might cause the request to fail. This edge-case is to cover up a possible scenario when responses.ResponseCompletedEvent is not sent

Thanks for the clarification, @hugoaguirre. I understand your concern that failing the request might be undesirable if the user has already received all the content chunks.

However, the ResponseCompletedEvent is crucial for populating the final ai.ModelResponse object, especially regarding usage statistics and ensuring the completeness of the final message. If this event is missing, the returned ai.ModelResponse will have an empty message and no usage data, which can lead to silent failures or misleading information for the caller.

An explicit error, as suggested, provides a clearer signal that the stream did not conclude as expected by the API, which is generally better for debugging and understanding unexpected API behavior. While the chunks might have been delivered, the overall contract of a completed response, including metadata, is not met without this event. This approach prioritizes explicit error handling over potentially ambiguous success.

/gemini review-comment-reply

Thanks for clarifying your preference to avoid outright failing the request if content chunks have already been sent, @hugoaguirre. I understand the desire to prioritize partial delivery over complete failure in such edge cases.

To address this while still providing clearer diagnostics to the caller, I suggest explicitly setting the FinishReason and FinishMessage fields in the ai.ModelResponse when the ResponseCompletedEvent is missing. This approach signals that the stream did not conclude as expected by the API (e.g., missing usage statistics or final metadata) without returning an error that might discard already processed chunks. This provides a more robust and informative response to the consumer of the API.

This modification would make the incomplete nature of the response explicit, allowing downstream logic to handle it appropriately, rather than implicitly assuming a complete, albeit empty, message.