Merge 1e5780b8dc into d266c59c2a

llama_stack/apis/vector_io/vector_io.py

@@ -92,6 +92,22 @@ class Chunk(BaseModel):
 
         return generate_chunk_id(str(uuid.uuid4()), str(self.content))
 
+    @property
+    def document_id(self) -> str | None:
+        """Returns the document_id from either metadata or chunk_metadata, with metadata taking precedence."""
+        # Check metadata first (takes precedence)
+        doc_id = self.metadata.get("document_id")
+        if isinstance(doc_id, str):
+            return doc_id
+
+        # Fall back to chunk_metadata if available
+        if self.chunk_metadata is not None:
+            chunk_doc_id = getattr(self.chunk_metadata, "document_id", None)
+            if isinstance(chunk_doc_id, str):
+                return chunk_doc_id
+
+        return None
+
 
 @json_schema_type
 class QueryChunksResponse(BaseModel):

llama_stack/core/routers/vector_io.py

@@ -5,6 +5,7 @@
 # the root directory of this source tree.
 
 import asyncio
+import logging  # allow-direct-logging
 import uuid
 from typing import Any
 
@@ -101,11 +102,14 @@ class VectorIORouter(VectorIO):
         chunks: list[Chunk],
         ttl_seconds: int | None = None,
     ) -> None:
-        logger.debug(
-            f"VectorIORouter.insert_chunks: {vector_db_id}, {len(chunks)} chunks, ttl_seconds={ttl_seconds}, chunk_ids={[chunk.metadata['document_id'] for chunk in chunks[:3]]}{' and more...' if len(chunks) > 3 else ''}",
-        )
+        if logger.isEnabledFor(logging.DEBUG):
+            doc_ids = [chunk.document_id for chunk in chunks[:3]]
+            logger.debug(
+                f"VectorIORouter.insert_chunks: {vector_db_id}, {len(chunks)} chunks, "
+                f"ttl_seconds={ttl_seconds}, chunk_ids={doc_ids}{' and more...' if len(chunks) > 3 else ''}"
+            )
         provider = await self.routing_table.get_provider_impl(vector_db_id)
-        return await provider.insert_chunks(vector_db_id, chunks, ttl_seconds)
+        await provider.insert_chunks(vector_db_id, chunks, ttl_seconds)
 
     async def query_chunks(
         self,

llama_stack/providers/inline/tool_runtime/rag/memory.py

@@ -278,7 +278,7 @@ class MemoryToolRuntimeImpl(ToolGroupsProtocolPrivate, ToolRuntime, RAGToolRunti
         return RAGQueryResult(
             content=picked,
             metadata={
-                "document_ids": [c.metadata["document_id"] for c in chunks[: len(picked)]],
+                "document_ids": [c.document_id for c in chunks[: len(picked)]],
                 "chunks": [c.content for c in chunks[: len(picked)]],
                 "scores": scores[: len(picked)],
                 "vector_db_ids": [c.metadata["vector_db_id"] for c in chunks[: len(picked)]],