feat: rebase and implement file API methods

Signed-off-by: Varsha Prasad Narsing <varshaprasad96@gmail.com>

llama_stack/providers/inline/vector_io/faiss/faiss.py

@@ -96,7 +96,7 @@ class FaissIndex(EmbeddingIndex):
 
         await self.kvstore.delete(f"{FAISS_INDEX_PREFIX}{self.bank_id}")
 
-    async def add_chunks(self, chunks: list[Chunk], embeddings: NDArray):
+    async def add_chunks(self, chunks: list[Chunk], embeddings: NDArray, metadata: dict[str, Any] | None = None):
         # Add dimension check
         embedding_dim = embeddings.shape[1] if len(embeddings.shape) > 1 else embeddings.shape[0]
         if embedding_dim != self.index.d:
@@ -234,6 +234,7 @@ class FaissVectorIOAdapter(OpenAIVectorStoreMixin, VectorIO, VectorDBsProtocolPr
         vector_db_id: str,
         chunks: list[Chunk],
         ttl_seconds: int | None = None,
+        params: dict[str, Any] | None = None,
     ) -> None:
         index = self.cache.get(vector_db_id)
         if index is None:

llama_stack/providers/inline/vector_io/qdrant/config.py

@@ -5,7 +5,7 @@
 # the root directory of this source tree.
 
 
-from typing import Any, Literal
+from typing import Any
 
 from pydantic import BaseModel
 
@@ -15,7 +15,6 @@ from llama_stack.schema_utils import json_schema_type
 @json_schema_type
 class QdrantVectorIOConfig(BaseModel):
     path: str
-    distance_metric: Literal["COSINE", "DOT", "EUCLID", "MANHATTAN"] = "COSINE"
 
     @classmethod
     def sample_run_config(cls, __distro_dir__: str) -> dict[str, Any]:

llama_stack/providers/inline/vector_io/sqlite_vec/sqlite_vec.py

@@ -178,7 +178,9 @@ class SQLiteVecIndex(EmbeddingIndex):
 
         await asyncio.to_thread(_drop_tables)
 
-    async def add_chunks(self, chunks: list[Chunk], embeddings: NDArray, batch_size: int = 500):
+    async def add_chunks(
+        self, chunks: list[Chunk], embeddings: NDArray, metadata: dict[str, Any] | None = None, batch_size: int = 500
+    ):
         """
         Add new chunks along with their embeddings using batch inserts.
         For each chunk, we insert its JSON into the metadata table and then insert its
@@ -729,7 +731,13 @@ class SQLiteVecVectorIOAdapter(OpenAIVectorStoreMixin, VectorIO, VectorDBsProtoc
 
         await asyncio.to_thread(_delete)
 
-    async def insert_chunks(self, vector_db_id: str, chunks: list[Chunk], ttl_seconds: int | None = None) -> None:
+    async def insert_chunks(
+        self,
+        vector_db_id: str,
+        chunks: list[Chunk],
+        ttl_seconds: int | None = None,
+        params: dict[str, Any] | None = None,
+    ) -> None:
         if vector_db_id not in self.cache:
             raise ValueError(f"Vector DB {vector_db_id} not found. Found: {list(self.cache.keys())}")
         # The VectorDBWithIndex helper is expected to compute embeddings via the inference_api