phoenix-oss/llama-stack-mirror
0
0
Fork 1
mirror of https://github.com/meta-llama/llama-stack.git synced 2025-12-21 09:48:41 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

chore(api)!: remove tool_runtime.rag_tool from the API surface (#4067)

Browse source 
RAG aka file search is implemented via the Responses API by specifying
the file-search tool. The backend implementation remains unchanged. This
PR merely removes the directly exposed API surface which allowed users
to directly perform searches from the client.

This facility is now available via the `client.vector_store.search()`
OpenAI compatible API.
This commit is contained in:
Ashwin Bharambe 2025-11-04 14:50:54 -08:00 • committed by GitHub
parent a8a8aa56c0
commit 0c49a53c97
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
10 changed files with 4 additions and 1117 deletions
Download patch file Download diff file Expand all files Collapse all files

40
src/llama_stack/core/routers/tool_runtime.py
View file

@ -8,14 +8,9 @@ from typing import Any
from llama_stack.apis.common.content_types import (
URL,
InterleavedContent,
)
from llama_stack.apis.tools import (
ListToolDefsResponse,
RAGDocument,
RAGQueryConfig,
RAGQueryResult,
RAGToolRuntime,
ToolRuntime,
)
from llama_stack.log import get_logger
@ -26,36 +21,6 @@ logger = get_logger(name=__name__, category="core::routers")
class ToolRuntimeRouter(ToolRuntime):
class RagToolImpl(RAGToolRuntime):
def __init__(
self,
routing_table: ToolGroupsRoutingTable,
) -> None:
logger.debug("Initializing ToolRuntimeRouter.RagToolImpl")
self.routing_table = routing_table
async def query(
self,
content: InterleavedContent,
vector_store_ids: list[str],
query_config: RAGQueryConfig | None = None,
) -> RAGQueryResult:
logger.debug(f"ToolRuntimeRouter.RagToolImpl.query: {vector_store_ids}")
provider = await self.routing_table.get_provider_impl("knowledge_search")
return await provider.query(content, vector_store_ids, query_config)
async def insert(
self,
documents: list[RAGDocument],
vector_store_id: str,
chunk_size_in_tokens: int = 512,
) -> None:
logger.debug(
f"ToolRuntimeRouter.RagToolImpl.insert: {vector_store_id}, {len(documents)} documents, chunk_size={chunk_size_in_tokens}"
)
provider = await self.routing_table.get_provider_impl("insert_into_memory")
return await provider.insert(documents, vector_store_id, chunk_size_in_tokens)
def __init__(
self,
routing_table: ToolGroupsRoutingTable,
@ -63,11 +28,6 @@ class ToolRuntimeRouter(ToolRuntime):
logger.debug("Initializing ToolRuntimeRouter")
self.routing_table = routing_table
# HACK ALERT this should be in sync with "get_all_api_endpoints()"
self.rag_tool = self.RagToolImpl(routing_table)
for method in ("query", "insert"):
setattr(self, f"rag_tool.{method}", getattr(self.rag_tool, method))
async def initialize(self) -> None:
logger.debug("ToolRuntimeRouter.initialize")
pass