Adding extra_headers parameters to ModelSettings

[jonnyk20]

Added the possibility to pass extra_headers when sending a request.
I used the same implementation as was recently used for extra_body and extra_query in #500

In this implementation I added the attributes to ModelSettings as suggested here
https://github.com/openai/openai-agents-python/issues/487 .

I'll be happy to add some tests if you have any suggestions.

[jonnyk20]

@rm-openai It doesn't look like I have the ability to request reviews. Is there a process I should follow?

[rm-openai]

Thanks @jonnyk20, looks good. Two change requests:

1. Can you please add tests? e.g. ensuring that extra_headers is passed on to the model if set.
2. Can you also add this to the litellm model?

Thanks1

[jonnyk20]

Thanks @jonnyk20, looks good. Two change requests:

1. Can you please add tests? e.g. ensuring that extra_headers is passed on to the model if set.
2. Can you also add this to the litellm model?

Thanks1

@rm-openai no problem - done! I wasn't able to create a LitellmModel test that didn't have exceptions, and I couldn't find any other tests using LitellmModel to use as reference. This is the closes I got:

@pytest.mark.allow_call_model_methods
@pytest.mark.asyncio
async def test_extra_headers_passed_to_litellm_model(monkeypatch):
    """
    Ensure extra_headers in ModelSettings is passed to the LitellmModel.
    """
    from agents.extensions.models.litellm_model import LitellmModel

    called_kwargs = {}

    async def dummy_acompletion(*args, **kwargs):
        nonlocal called_kwargs
        called_kwargs = kwargs
        return None

    monkeypatch.setattr("agents.extensions.models.litellm_model.litellm.acompletion", dummy_acompletion)

    model = LitellmModel(model="any-model")
    extra_headers = {"X-Test-Header": "test-value"}
    
    await model.get_response(
        system_instructions=None,
        input="hi",
        model_settings=ModelSettings(extra_headers=extra_headers),
        tools=[],
        output_schema=None,
        handoffs=[],
        tracing=ModelTracing.DISABLED,
        previous_response_id=None,
    )
    assert "extra_headers" in called_kwargs
    assert called_kwargs["extra_headers"]["X-Test-Header"] == "test-value"

Which resulted in the following:

=================================== FAILURES ===================================
__________________ test_extra_headers_passed_to_litellm_model __________________

monkeypatch = <_pytest.monkeypatch.MonkeyPatch object at 0x119490730>

    @pytest.mark.allow_call_model_methods
    @pytest.mark.asyncio
    async def test_extra_headers_passed_to_litellm_model(monkeypatch):
        """
        Ensure extra_headers in ModelSettings is passed to the LitellmModel.
        """
        from agents.extensions.models.litellm_model import LitellmModel
    
        called_kwargs = {}
    
        async def dummy_acompletion(*args, **kwargs):
            nonlocal called_kwargs
            called_kwargs = kwargs
            return None
    
        monkeypatch.setattr("agents.extensions.models.litellm_model.litellm.acompletion", dummy_acompletion)
    
        model = LitellmModel(model="any-model")
        extra_headers = {"X-Test-Header": "test-value"}
    
>       await model.get_response(
            system_instructions=None,
            input="hi",
            model_settings=ModelSettings(extra_headers=extra_headers),
            tools=[],
            output_schema=None,
            handoffs=[],
            tracing=ModelTracing.DISABLED,
            previous_response_id=None,
        )

tests/test_extra_headers.py:116: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

self = <agents.extensions.models.litellm_model.LitellmModel object at 0x119490a30>
system_instructions = None, input = 'hi'
model_settings = ModelSettings(temperature=None, top_p=None, frequency_penalty=None, presence_penalty=None, tool_choice=None, parallel_...None, store=None, include_usage=None, extra_query=None, extra_body=None, extra_headers={'X-Test-Header': 'test-value'})
tools = [], output_schema = None, handoffs = []
tracing = <ModelTracing.DISABLED: 0>, previous_response_id = None

    async def get_response(
        self,
        system_instructions: str | None,
        input: str | list[TResponseInputItem],
        model_settings: ModelSettings,
        tools: list[Tool],
        output_schema: AgentOutputSchema | None,
        handoffs: list[Handoff],
        tracing: ModelTracing,
        previous_response_id: str | None,
    ) -> ModelResponse:
        with generation_span(
            model=str(self.model),
            model_config=dataclasses.asdict(model_settings)
            | {"base_url": str(self.base_url or ""), "model_impl": "litellm"},
            disabled=tracing.is_disabled(),
        ) as span_generation:
            response = await self._fetch_response(
                system_instructions,
                input,
                model_settings,
                tools,
                output_schema,
                handoffs,
                span_generation,
                tracing,
                stream=False,
            )
    
>           assert isinstance(response.choices[0], litellm.types.utils.Choices)
E           AttributeError: 'tuple' object has no attribute 'choices'

src/agents/extensions/models/litellm_model.py:94: AttributeError
=============================== warnings summary ===============================
tests/test_extra_headers.py::test_extra_headers_passed_to_litellm_model
  /Users/jonnykalambay/projects/oss/openai-agents-python/.venv/lib/python3.9/site-packages/pydantic/_internal/_config.py:323: PydanticDeprecatedSince20: Support for class-based `config` is deprecated, use ConfigDict instead. Deprecated in Pydantic V2.0 to be removed in V3.0. See Pydantic V2 Migration Guide at https://errors.pydantic.dev/2.11/migration/
    warnings.warn(DEPRECATION_MESSAGE, DeprecationWarning)

-- Docs: https://docs.pytest.org/en/stable/how-to/capture-warnings.html
=========================== short test summary info ============================
FAILED tests/test_extra_headers.py::test_extra_headers_passed_to_litellm_model - AttributeError: 'tuple' object has no attribute 'choices'
==================== 1 failed, 2 passed, 1 warning in 1.08s ====================
(openai-agents) ➜  openai-agents-python git:(add_extra_headers_parameters) ✗ 

Any suggestions for how to fix it?

[rm-openai]

@jonnyk20 PR looks good, would you mind fixing lint and I can merge?

[jonnyk20]

@jonnyk20 PR looks good, would you mind fixing lint and I can merge?

@rm-openai, sure done. Thanks for the review!