fix: UnboundLocalError for role in streaming audio/image responses (#784)

Signed-off-by: Pierre Le Guen <26087574+PierreLeGuen@users.noreply.github.com>

vllm_omni/entrypoints/openai/serving_chat.py

@@ -667,17 +667,17 @@ class OmniOpenAIServingChat(OpenAIServingChat, AudioMixin):
                     if res.encoder_prompt_token_ids is not None:
                         num_prompt_tokens += len(res.encoder_prompt_token_ids)
 
+                # Initialize role before conditional blocks to avoid UnboundLocalError
+                # when handling audio/image responses
+                role = self.get_chat_request_role(request)
+
                 # We need to do it here, because if there are exceptions in
                 # the result_generator, it needs to be sent as the FIRST
                 # response (by the try...catch).
                 if first_iteration_dict[final_output_type] and final_output_type == "text":
                     num_cached_tokens = res.num_cached_tokens
-                    # Send first response for each request.n (index) with
-                    # the role
-                    role = self.get_chat_request_role(request)
-
-                    # NOTE num_choices defaults to 1 so this usually executes
-                    # once per request
+                    # Send first response for each choice with role
+                    # NOTE: num_choices defaults to 1 so this usually executes once per request
                     for i in range(num_choices):
                         choice_data = ChatCompletionResponseStreamChoice(
                             index=i,