Python: Bug: Assistant Agent enable json error

[onderyildirim]

Describe the bug
Wrong option set when enable_json_response=True

To Reproduce
Steps to reproduce the behavior:
a = await AzureAssistantAgent.create(
kernel=Kernel(),
service_id="agent",
name="a1",
instructions="",
enable_json_response=True,
)
tid = await a.create_thread()
await a.add_chat_message(thread_id=tid,message=ChatMessageContent(role=AuthorRole.USER, content="Return numbers 1 to 10 in json format"))
async for r in a.invoke(thread_id=tid):
ar = r.content
print(ar)

invoke method fails with
File "...\main.py", line 357, in main
async for response in production_optimizer_agent.invoke_stream(thread_id=optimizer_thread_id):
File "....venv\Lib\site-packages\semantic_kernel\agents\open_ai\open_ai_assistant_base.py", line 827, in invoke_stream
async for content in self._invoke_internal_stream(
File "....venv\Lib\site-packages\semantic_kernel\agents\open_ai\open_ai_assistant_base.py", line 905, in _invoke_internal_stream
async with stream as response_stream:
^^^^^^
File "....venv\Lib\site-packages\openai\lib\streaming_assistants.py", line 879, in aenter
self.__stream = await self.__api_request
^^^^^^^^^^^^^^^^^^^^^^^^
File "....venv\Lib\site-packages\openai_base_client.py", line 1839, in post
return await self.request(cast_to, opts, stream=stream, stream_cls=stream_cls)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "....venv\Lib\site-packages\openai_base_client.py", line 1533, in request
return await self._request(
^^^^^^^^^^^^^^^^^^^^
File "....venv\Lib\site-packages\openai_base_client.py", line 1634, in _request
raise self._make_status_error_from_response(err.response) from None
openai.BadRequestError: Error code: 400 - {'error': {'message': "Invalid value: 'json'. Supported values are: 'auto'.", 'type': 'invalid_request_error', 'param': 'response_format', 'code': 'invalid_value'}}

Expected behavior
A response like below
{
"numbers": [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
}

Platform

• Language: Python

Changing ....venv\Lib\site-packages\semantic_kernel\agents\open_ai\open_ai_assistant_base.py line 1052 (method _generate_options)

from:
"response_format": "json" if merged_options.get("enable_json_response") else None,
to:
"response_format": {"type": "json_object"} if merged_options.get("enable_json_response") else None,

solves the problem.

[shethaadit]

Hi @markwallace-microsoft / @moonbox3, I will raise a quick PR for the fix. :)