diff --git a/litellm/main.py b/litellm/main.py index c22cd8eb11c..78fb6dcb11b 100644 --- a/litellm/main.py +++ b/litellm/main.py @@ -7961,8 +7961,11 @@ def text_completion( # Support for token IDs as prompt (list of integers or list of lists of integers) messages = [{"role": "user", "content": prompt}] else: - raise Exception( - f"Unmapped prompt format. Your prompt is neither a list of strings nor a string. prompt={prompt}. File an issue - https://github.com/BerriAI/litellm/issues" + raise litellm.BadRequestError( + message=f"Unmapped prompt format. Your prompt is neither a list of strings nor a string. prompt={prompt}. File an issue - https://github.com/BerriAI/litellm/issues", + model=model or "", + llm_provider=custom_llm_provider or "", + body={"type": "invalid_request_error", "param": "prompt"}, ) kwargs.pop("prompt", None) diff --git a/litellm/utils.py b/litellm/utils.py index 1db5f7129df..a4baf022f03 100644 --- a/litellm/utils.py +++ b/litellm/utils.py @@ -8983,8 +8983,11 @@ def validate_chat_completion_user_messages(messages: list[AllMessageValues]): f"Invalid message at index {idx}. Please ensure all messages are valid OpenAI chat completion messages." ) if "invalid content type" in str(e): - raise Exception( - f"Invalid user message at index {idx}. Please ensure all user messages are valid OpenAI chat completion messages." + raise BadRequestError( + message=f"Invalid user message at index {idx}. Please ensure all user messages are valid OpenAI chat completion messages.", + model="", + llm_provider="", + body={"type": "invalid_request_error", "param": f"messages[{idx}].content"}, ) else: raise e diff --git a/tests/unit/test_main.py b/tests/unit/test_main.py index 70b268bf2a2..e6e6c89627f 100644 --- a/tests/unit/test_main.py +++ b/tests/unit/test_main.py @@ -9950,3 +9950,54 @@ async def test_converse_fallback_prices_the_current_attempt( else: client.close() configuration.reset_rust_configuration() + + +@pytest.mark.parametrize("asynchronous", [False, True]) +@pytest.mark.asyncio +async def test_unknown_chat_content_is_a_non_retryable_bad_request(asynchronous: bool) -> None: + from litellm.proxy.common_request_processing import proxy_exception_from_route_error + + messages: Final = [{"role": "user", "content": [{"type": "text", "text": "hi"}, {"type": "future_block"}]}] + if asynchronous: + with pytest.raises(litellm.BadRequestError, match="Invalid user message at index 0") as raised: + await litellm.acompletion(model="hosted_vllm/local-qa", messages=messages, api_key="local-fixture") + else: + with pytest.raises(litellm.BadRequestError, match="Invalid user message at index 0") as raised: + litellm.completion(model="hosted_vllm/local-qa", messages=messages, api_key="local-fixture") + + assert raised.value.status_code == 400 + assert raised.value.type == "invalid_request_error" + assert raised.value.param == "messages[0].content" + assert not litellm.should_retry(raised.value.status_code) + mapped: Final = proxy_exception_from_route_error(raised.value) + assert mapped.type == "invalid_request_error" + assert mapped.code == "400" + with pytest.raises(litellm.BadRequestError) as router_error: + litellm.Router().should_retry_this_error(raised.value, healthy_deployments=[{}]) + assert router_error.value is raised.value + + +@pytest.mark.parametrize("asynchronous", [False, True]) +@pytest.mark.asyncio +async def test_object_text_prompt_is_a_non_retryable_bad_request(asynchronous: bool) -> None: + from litellm.proxy.common_request_processing import proxy_exception_from_route_error + + if asynchronous: + with pytest.raises(litellm.BadRequestError, match="Unmapped prompt format") as raised: + await litellm.atext_completion( + model="hosted_vllm/local-qa", prompt={"not": "a prompt"}, api_key="local-fixture" + ) + else: + with pytest.raises(litellm.BadRequestError, match="Unmapped prompt format") as raised: + litellm.text_completion(model="hosted_vllm/local-qa", prompt={"not": "a prompt"}, api_key="local-fixture") + + assert raised.value.status_code == 400 + assert raised.value.type == "invalid_request_error" + assert raised.value.param == "prompt" + assert not litellm.should_retry(raised.value.status_code) + mapped: Final = proxy_exception_from_route_error(raised.value) + assert mapped.type == "invalid_request_error" + assert mapped.code == "400" + with pytest.raises(litellm.BadRequestError) as router_error: + litellm.Router().should_retry_this_error(raised.value, healthy_deployments=[{}]) + assert router_error.value is raised.value