From 2d6af6a7b1fa14a227a688293c4289bce6584a76 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E3=81=B5=E3=81=81=E3=83=BC?=
 <47295014+ymuichiro@users.noreply.github.com>
Date: Tue, 19 Nov 2024 18:09:52 +0900
Subject: [PATCH] Python: Introduced a new condition to yield
 `StreamingChatMessageContent` directly when usage data is available.

---
 .../open_ai/services/azure_chat_completion.py | 26 ++++++++++++++++---
 1 file changed, 22 insertions(+), 4 deletions(-)

diff --git a/python/semantic_kernel/connectors/ai/open_ai/services/azure_chat_completion.py b/python/semantic_kernel/connectors/ai/open_ai/services/azure_chat_completion.py
index bd2a0ca51bab..6152f0eae3ae 100644
--- a/python/semantic_kernel/connectors/ai/open_ai/services/azure_chat_completion.py
+++ b/python/semantic_kernel/connectors/ai/open_ai/services/azure_chat_completion.py
@@ -8,6 +8,8 @@
 from typing import Any, TypeVar
 from uuid import uuid4
 
+from semantic_kernel.contents import AuthorRole
+
 if sys.version_info >= (3, 12):
     from typing import override  # pragma: no cover
 else:
@@ -148,14 +150,30 @@ async def _inner_get_streaming_chat_message_contents(
         if not isinstance(response, AsyncStream):
             raise ServiceInvalidResponseError("Expected an AsyncStream[ChatCompletionChunk] response.")
         async for chunk in response:
-            if len(chunk.choices) == 0:
+            if len(chunk.choices) == 0 and chunk.usage is None:
                 continue
 
             assert isinstance(chunk, ChatCompletionChunk)  # nosec
             chunk_metadata = self._get_metadata_from_streaming_chat_response(chunk)
-            yield [
-                self._create_streaming_chat_message_content(chunk, choice, chunk_metadata) for choice in chunk.choices
-            ]
+
+            if chunk.usage is not None:
+                yield [
+                    StreamingChatMessageContent(
+                        role=AuthorRole.ASSISTANT,
+                        content="",
+                        choice_index=i,
+                        inner_content=chunk,
+                        ai_model_id=settings.ai_model_id,
+                        metadata=chunk_metadata,
+                    )
+                    for i in range(settings.number_of_responses or 1)
+                ]
+
+            else:
+                yield [
+                    self._create_streaming_chat_message_content(chunk, choice, chunk_metadata)
+                    for choice in chunk.choices
+                ]
 
     @classmethod
     def from_dict(cls, settings: dict[str, Any]) -> "AzureChatCompletion":