diff --git a/litellm/interactions/litellm_responses_transformation/transformation.py b/litellm/interactions/litellm_responses_transformation/transformation.py index 2a71c3e8977..4209f1538e0 100644 --- a/litellm/interactions/litellm_responses_transformation/transformation.py +++ b/litellm/interactions/litellm_responses_transformation/transformation.py @@ -2,10 +2,11 @@ Transformation utilities for bridging Interactions API to Responses API. This module handles transforming between: -- Interactions API format (Google's format with Turn[], system_instruction, etc.) +- Interactions API format (Google's format with Step[]/Turn[], system_instruction, etc.) - Responses API format (OpenAI's format with input[], instructions, etc.) """ +from types import MappingProxyType from typing import Any, Final, cast from litellm.types.interactions import ( @@ -19,6 +20,8 @@ from litellm.types.llms.openai import ( ResponsesAPIResponse, ) +_STEP_TYPE_ROLES: Final = MappingProxyType({"user_input": "user", "model_output": "assistant"}) + class LiteLLMResponsesInteractionsConfig: """Configuration class for transforming between Interactions API and Responses API.""" @@ -91,112 +94,98 @@ class LiteLLMResponsesInteractionsConfig: Interactions API input can be: - string: "Hello" - - Turn[]: [{"role": "user", "content": [...]}] - - Content object + - Step[]: [{"type": "user_input", "content": [...]}, {"type": "model_output", "content": [...]}] + - Turn[] (legacy): [{"role": "user", "content": [...]}] + - Content | Content[]: one user message worth of content parts Responses API input is: - string: "Hello" - - Message[]: [{"role": "user", "content": [...]}] + - Message[]: [{"role": "user", "content": [{"type": "input_text", ...}]}] """ if isinstance(input, str): - # ResponseInputParam accepts str return cast(ResponseInputParam, input) if isinstance(input, list): - # Turn[] format - convert to Responses API Message[] format - messages: Final = [] - for turn in input: - if isinstance(turn, dict): - role = turn.get("role", "user") - content = turn.get("content", []) - - # Transform content array - transformed_content = LiteLLMResponsesInteractionsConfig._transform_content_array(content) - - messages.append( - { - "role": role, - "content": transformed_content, - } - ) - elif isinstance(turn, Turn): - # Pydantic model - role = turn.role if hasattr(turn, "role") else "user" - content = turn.content if hasattr(turn, "content") else [] - - # Ensure content is a list for _transform_content_array - # Cast to List[Any] to handle various content types - if isinstance(content, list): - content_list: list[Any] = list(content) - elif content is not None: - content_list = [content] - else: - content_list = [] - - transformed_content = LiteLLMResponsesInteractionsConfig._transform_content_array(content_list) - - messages.append( - { - "role": role, - "content": transformed_content, - } - ) - - return cast(ResponseInputParam, messages) - - # Single content object - wrap in message - if isinstance(input, dict): + if any(LiteLLMResponsesInteractionsConfig._is_history_item(item) for item in input): + return cast( + ResponseInputParam, + [ + LiteLLMResponsesInteractionsConfig._transform_history_item(item) + for item in input + if LiteLLMResponsesInteractionsConfig._is_history_item(item) + ], + ) return cast( ResponseInputParam, [ { "role": "user", - "content": LiteLLMResponsesInteractionsConfig._transform_content_array( - input.get("content", []) if isinstance(input.get("content"), list) else [input] - ), + "content": LiteLLMResponsesInteractionsConfig._transform_content_array(list(input), "user"), + } + ], + ) + + if isinstance(input, dict): + raw_content: Final = input.get("content") + content_items: Final = raw_content if isinstance(raw_content, list) else [input] + return cast( + ResponseInputParam, + [ + { + "role": "user", + "content": LiteLLMResponsesInteractionsConfig._transform_content_array(content_items, "user"), } ], ) - # Fallback: convert to string return cast(ResponseInputParam, str(input)) @staticmethod - def _transform_content_array(content: list[Any]) -> list[dict[str, Any]]: - """Transform Interactions API content array to Responses API format.""" - if not isinstance(content, list): - # Single content item - wrap in array - content = [content] + def _is_history_item(item: Any) -> bool: + if isinstance(item, Turn): + return True + return isinstance(item, dict) and ("role" in item or item.get("type") in _STEP_TYPE_ROLES) - transformed: Final[list[dict[str, Any]]] = [] - for item in content: - if isinstance(item, dict): - # Already in dict format, pass through - transformed.append(item) - elif isinstance(item, str): - # Plain string - wrap in text format - transformed.append({"type": "text", "text": item}) - else: - # Pydantic model or other - convert to dict - if hasattr(item, "model_dump"): - dumped = item.model_dump() - if isinstance(dumped, dict): - transformed.append(dumped) - else: - # Fallback: wrap in text format - transformed.append({"type": "text", "text": str(dumped)}) - elif hasattr(item, "dict"): - dumped = item.dict() - if isinstance(dumped, dict): - transformed.append(dumped) - else: - # Fallback: wrap in text format - transformed.append({"type": "text", "text": str(dumped)}) - else: - # Fallback: wrap in text format - transformed.append({"type": "text", "text": str(item)}) + @staticmethod + def _transform_history_item(item: "Turn | dict[str, Any]") -> dict[str, Any]: + raw: Final = item.model_dump(exclude_none=True) if isinstance(item, Turn) else item + role: Final = LiteLLMResponsesInteractionsConfig._responses_role(raw) + raw_content: Final = raw.get("content") + content_items: Final = ( + raw_content if isinstance(raw_content, list) else [] if raw_content is None else [raw_content] + ) + return { + "role": role, + "content": LiteLLMResponsesInteractionsConfig._transform_content_array(content_items, role), + } - return transformed + @staticmethod + def _responses_role(item: dict[str, Any]) -> str: + step_role: Final = _STEP_TYPE_ROLES.get(str(item.get("type", ""))) + if step_role is not None: + return step_role + raw_role: Final = str(item.get("role") or "user") + return "assistant" if raw_role == "model" else raw_role + + @staticmethod + def _transform_content_array(content: list[Any], role: str) -> list[dict[str, Any]]: + """Transform Interactions API content parts to Responses API parts for the given role.""" + return [LiteLLMResponsesInteractionsConfig._transform_content_item(item, role) for item in content] + + @staticmethod + def _transform_content_item(item: Any, role: str) -> dict[str, Any]: + text_type: Final = "output_text" if role == "assistant" else "input_text" + if isinstance(item, str): + return {"type": text_type, "text": item} + if isinstance(item, dict): + if item.get("type") == "text": + return {"type": text_type, "text": str(item.get("text", ""))} + return item + if hasattr(item, "model_dump"): + dumped: Final = item.model_dump(exclude_none=True) + if isinstance(dumped, dict): + return LiteLLMResponsesInteractionsConfig._transform_content_item(dumped, role) + return {"type": text_type, "text": str(item)} @staticmethod def transform_responses_response_to_interactions_response( diff --git a/tests/test_litellm/interactions/test_litellm_responses_bridge.py b/tests/test_litellm/interactions/test_litellm_responses_bridge.py index 17e7f9fc4ff..8400f2c4840 100644 --- a/tests/test_litellm/interactions/test_litellm_responses_bridge.py +++ b/tests/test_litellm/interactions/test_litellm_responses_bridge.py @@ -7,6 +7,10 @@ the litellm_responses bridge provider, which calls litellm.responses() internall import os +from litellm.interactions.litellm_responses_transformation.transformation import ( + LiteLLMResponsesInteractionsConfig, +) +from litellm.types.interactions import Turn from tests.test_litellm.interactions.base_interactions_test import ( BaseInteractionsTest, ) @@ -26,3 +30,71 @@ class TestLiteLLMResponsesBridge(BaseInteractionsTest): def get_api_key(self) -> str: """Return the OpenAI API key from environment.""" return os.getenv("OPENAI_API_KEY", "") + + +class TestBridgeInputTransformation: + """Regression tests for translating Interactions input into Responses API input. + + The bridge used to pass Google content parts through raw ({"type": "text"}), + which the Responses API rejects with a 400, and it dropped the role encoded + in step types and in the legacy "model" turn role. + """ + + def test_step_input_maps_roles_and_content_types(self): + transformed = LiteLLMResponsesInteractionsConfig._transform_interactions_input_to_responses_input( + [ + {"type": "user_input", "content": [{"type": "text", "text": "I like apples."}]}, + {"type": "model_output", "content": [{"type": "text", "text": "I like oranges."}]}, + {"type": "user_input", "content": [{"type": "text", "text": "What did you say?"}]}, + ] + ) + assert transformed == [ + {"role": "user", "content": [{"type": "input_text", "text": "I like apples."}]}, + {"role": "assistant", "content": [{"type": "output_text", "text": "I like oranges."}]}, + {"role": "user", "content": [{"type": "input_text", "text": "What did you say?"}]}, + ] + + def test_legacy_turn_input_maps_model_role_to_assistant(self): + transformed = LiteLLMResponsesInteractionsConfig._transform_interactions_input_to_responses_input( + [ + {"role": "user", "content": [{"type": "text", "text": "I like apples."}]}, + {"role": "model", "content": [{"type": "text", "text": "I like oranges."}]}, + ] + ) + assert transformed == [ + {"role": "user", "content": [{"type": "input_text", "text": "I like apples."}]}, + {"role": "assistant", "content": [{"type": "output_text", "text": "I like oranges."}]}, + ] + + def test_turn_pydantic_model_with_string_content(self): + transformed = LiteLLMResponsesInteractionsConfig._transform_interactions_input_to_responses_input( + [Turn(role="model", content="I like oranges.")] + ) + assert transformed == [ + {"role": "assistant", "content": [{"type": "output_text", "text": "I like oranges."}]} + ] + + def test_string_input_passes_through(self): + transformed = LiteLLMResponsesInteractionsConfig._transform_interactions_input_to_responses_input("Hello") + assert transformed == "Hello" + + def test_content_list_input_becomes_single_user_message(self): + transformed = LiteLLMResponsesInteractionsConfig._transform_interactions_input_to_responses_input( + [{"type": "text", "text": "Hello"}, "world"] + ) + assert transformed == [ + { + "role": "user", + "content": [ + {"type": "input_text", "text": "Hello"}, + {"type": "input_text", "text": "world"}, + ], + } + ] + + def test_non_text_content_passes_through_unchanged(self): + image_part = {"type": "image", "data": "base64data", "mime_type": "image/png"} + transformed = LiteLLMResponsesInteractionsConfig._transform_interactions_input_to_responses_input( + [{"type": "user_input", "content": [image_part]}] + ) + assert transformed == [{"role": "user", "content": [image_part]}]