phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-12-29 23:09:32 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

fold openai responses into the Agents API

Browse source
This commit is contained in:
Ashwin Bharambe 2025-04-28 10:27:28 -07:00
parent 207224a811
commit abd6280cb8
25 changed files with 967 additions and 199 deletions
Download patch file Download diff file Expand all files Collapse all files

31
llama_stack/providers/inline/agents/meta_reference/agents.py
View file

@ -23,6 +23,9 @@ from llama_stack.apis.agents import (
Document,
ListAgentSessionsResponse,
ListAgentsResponse,
OpenAIResponseInputMessage,
OpenAIResponseInputTool,
OpenAIResponseObject,
Session,
Turn,
)
@ -40,6 +43,7 @@ from llama_stack.providers.utils.kvstore import InmemoryKVStoreImpl, kvstore_imp
from .agent_instance import ChatAgent
from .config import MetaReferenceAgentsImplConfig
from .openai_responses import OpenAIResponsesImpl
logger = logging.getLogger()
logger.setLevel(logging.INFO)
@ -63,9 +67,16 @@ class MetaReferenceAgentsImpl(Agents):
self.tool_groups_api = tool_groups_api
self.in_memory_store = InmemoryKVStoreImpl()
self.openai_responses_impl = None
async def initialize(self) -> None:
self.persistence_store = await kvstore_impl(self.config.persistence_store)
self.openai_responses_impl = OpenAIResponsesImpl(
self.persistence_store,
inference_api=self.inference_api,
tool_groups_api=self.tool_groups_api,
tool_runtime_api=self.tool_runtime_api,
)
# check if "bwrap" is available
if not shutil.which("bwrap"):
@ -244,3 +255,23 @@ class MetaReferenceAgentsImpl(Agents):
agent_id: str,
) -> ListAgentSessionsResponse:
pass
# OpenAI responses
async def get_openai_response(
self,
id: str,
) -> OpenAIResponseObject:
return await self.openai_responses_impl.get_openai_response(id)
async def create_openai_response(
self,
input: Union[str, List[OpenAIResponseInputMessage]],
model: str,
previous_response_id: Optional[str] = None,
store: Optional[bool] = True,
stream: Optional[bool] = False,
tools: Optional[List[OpenAIResponseInputTool]] = None,
) -> OpenAIResponseObject:
return await self.openai_responses_impl.create_openai_response(
input, model, previous_response_id, store, stream, tools
)

64
llama_stack/providers/inline/openai_responses/openai_responses.py → llama_stack/providers/inline/agents/meta_reference/openai_responses.py
View file

@ -10,6 +10,20 @@ from typing import AsyncIterator, List, Optional, Union, cast
from openai.types.chat import ChatCompletionToolParam
from llama_stack.apis.agents.openai_responses import (
OpenAIResponseInputMessage,
OpenAIResponseInputMessageContentImage,
OpenAIResponseInputMessageContentText,
OpenAIResponseInputTool,
OpenAIResponseObject,
OpenAIResponseObjectStream,
OpenAIResponseObjectStreamResponseCompleted,
OpenAIResponseObjectStreamResponseCreated,
OpenAIResponseOutput,
OpenAIResponseOutputMessage,
OpenAIResponseOutputMessageContentOutputText,
OpenAIResponseOutputMessageWebSearchToolCall,
)
from llama_stack.apis.inference.inference import (
Inference,
OpenAIAssistantMessageParam,
@ -24,29 +38,11 @@ from llama_stack.apis.inference.inference import (
OpenAIToolMessageParam,
OpenAIUserMessageParam,
)
from llama_stack.apis.models.models import Models, ModelType
from llama_stack.apis.openai_responses import OpenAIResponses
from llama_stack.apis.openai_responses.openai_responses import (
OpenAIResponseInputMessage,
OpenAIResponseInputMessageContentImage,
OpenAIResponseInputMessageContentText,
OpenAIResponseInputTool,
OpenAIResponseObject,
OpenAIResponseObjectStream,
OpenAIResponseObjectStreamResponseCompleted,
OpenAIResponseObjectStreamResponseCreated,
OpenAIResponseOutput,
OpenAIResponseOutputMessage,
OpenAIResponseOutputMessageContentOutputText,
OpenAIResponseOutputMessageWebSearchToolCall,
)
from llama_stack.apis.tools.tools import ToolGroups, ToolInvocationResult, ToolRuntime
from llama_stack.log import get_logger
from llama_stack.models.llama.datatypes import ToolDefinition, ToolParamDefinition
from llama_stack.providers.utils.inference.openai_compat import convert_tooldef_to_openai_tool
from llama_stack.providers.utils.kvstore import kvstore_impl
from .config import OpenAIResponsesImplConfig
from llama_stack.providers.utils.kvstore import KVStore
logger = get_logger(name=__name__, category="openai_responses")
@ -80,34 +76,25 @@ async def _openai_choices_to_output_messages(choices: List[OpenAIChoice]) -> Lis
return output_messages
class OpenAIResponsesImpl(OpenAIResponses):
class OpenAIResponsesImpl:
def __init__(
self,
config: OpenAIResponsesImplConfig,
models_api: Models,
persistence_store: KVStore,
inference_api: Inference,
tool_groups_api: ToolGroups,
tool_runtime_api: ToolRuntime,
):
self.config = config
self.models_api = models_api
self.persistence_store = persistence_store
self.inference_api = inference_api
self.tool_groups_api = tool_groups_api
self.tool_runtime_api = tool_runtime_api
async def initialize(self) -> None:
self.kvstore = await kvstore_impl(self.config.kvstore)
async def shutdown(self) -> None:
logger.debug("OpenAIResponsesImpl.shutdown")
pass
async def get_openai_response(
self,
id: str,
) -> OpenAIResponseObject:
key = f"{OPENAI_RESPONSES_PREFIX}{id}"
response_json = await self.kvstore.get(key=key)
response_json = await self.persistence_store.get(key=key)
if response_json is None:
raise ValueError(f"OpenAI response with id '{id}' not found")
return OpenAIResponseObject.model_validate_json(response_json)
@ -122,11 +109,6 @@ class OpenAIResponsesImpl(OpenAIResponses):
tools: Optional[List[OpenAIResponseInputTool]] = None,
):
stream = False if stream is None else stream
model_obj = await self.models_api.get_model(model)
if model_obj is None:
raise ValueError(f"Model '{model}' not found")
if model_obj.model_type == ModelType.embedding:
raise ValueError(f"Model '{model}' is an embedding model and does not support chat completions")
messages: List[OpenAIMessageParam] = []
if previous_response_id:
@ -155,7 +137,7 @@ class OpenAIResponsesImpl(OpenAIResponses):
chat_tools = await self._convert_response_tools_to_chat_tools(tools) if tools else None
chat_response = await self.inference_api.openai_chat_completion(
model=model_obj.identifier,
model=model,
messages=messages,
tools=chat_tools,
stream=stream,
@ -198,14 +180,14 @@ class OpenAIResponsesImpl(OpenAIResponses):
output_messages: List[OpenAIResponseOutput] = []
if chat_response.choices[0].finish_reason == "tool_calls":
output_messages.extend(
await self._execute_tool_and_return_final_output(model_obj.identifier, stream, chat_response, messages)
await self._execute_tool_and_return_final_output(model, stream, chat_response, messages)
)
else:
output_messages.extend(await _openai_choices_to_output_messages(chat_response.choices))
response = OpenAIResponseObject(
created_at=chat_response.created,
id=f"resp-{uuid.uuid4()}",
model=model_obj.identifier,
model=model,
object="response",
status="completed",
output=output_messages,
@ -214,7 +196,7 @@ class OpenAIResponsesImpl(OpenAIResponses):
if store:
# Store in kvstore
key = f"{OPENAI_RESPONSES_PREFIX}{response.id}"
await self.kvstore.set(
await self.persistence_store.set(
key=key,
value=response.model_dump_json(),
)

21
llama_stack/providers/inline/openai_responses/__init__.py
View file

@ -1,21 +0,0 @@
# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the terms described in the LICENSE file in
# the root directory of this source tree.
from typing import Any, Dict
from llama_stack.apis.datatypes import Api
from .config import OpenAIResponsesImplConfig
async def get_provider_impl(config: OpenAIResponsesImplConfig, deps: Dict[Api, Any]):
from .openai_responses import OpenAIResponsesImpl
impl = OpenAIResponsesImpl(
config, deps[Api.models], deps[Api.inference], deps[Api.tool_groups], deps[Api.tool_runtime]
)
await impl.initialize()
return impl

24
llama_stack/providers/inline/openai_responses/config.py
View file

@ -1,24 +0,0 @@
# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the terms described in the LICENSE file in
# the root directory of this source tree.
from typing import Any, Dict
from pydantic import BaseModel
from llama_stack.providers.utils.kvstore.config import KVStoreConfig, SqliteKVStoreConfig
class OpenAIResponsesImplConfig(BaseModel):
kvstore: KVStoreConfig
@classmethod
def sample_run_config(cls, __distro_dir__: str, **kwargs: Any) -> Dict[str, Any]:
return {
"kvstore": SqliteKVStoreConfig.sample_run_config(
__distro_dir__=__distro_dir__,
db_name="openai_responses.db",
)
}

27
llama_stack/providers/registry/openai_responses.py
View file

@ -1,27 +0,0 @@
# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the terms described in the LICENSE file in
# the root directory of this source tree.
from typing import List
from llama_stack.providers.datatypes import Api, InlineProviderSpec, ProviderSpec
def available_providers() -> List[ProviderSpec]:
return [
InlineProviderSpec(
api=Api.openai_responses,
provider_type="inline::openai-responses",
pip_packages=[],
module="llama_stack.providers.inline.openai_responses",
config_class="llama_stack.providers.inline.openai_responses.config.OpenAIResponsesImplConfig",
api_dependencies=[
Api.models,
Api.inference,
Api.tool_groups,
Api.tool_runtime,
],
),
]