feat: add support in vector_store for server-side embeddings (#22)

* Adds vectorize functionality

libs/astradb/langchain_astradb/__init__.py

@@ -1,3 +1,5 @@
+from astrapy.info import CollectionVectorServiceOptions
+
 from langchain_astradb.cache import AstraDBCache, AstraDBSemanticCache
 from langchain_astradb.chat_message_histories import AstraDBChatMessageHistory
 from langchain_astradb.document_loaders import AstraDBLoader
@@ -12,4 +14,5 @@
     "AstraDBChatMessageHistory",
     "AstraDBLoader",
     "AstraDBVectorStore",
+    "CollectionVectorServiceOptions",
 ]

libs/astradb/langchain_astradb/utils/astradb.py

@@ -11,6 +11,7 @@
 import langchain_core
 from astrapy.api import APIRequestError
 from astrapy.db import AstraDB, AstraDBCollection, AsyncAstraDB, AsyncAstraDBCollection
+from astrapy.info import CollectionVectorServiceOptions
 
 
 class SetupMode(Enum):
@@ -94,6 +95,9 @@ def __init__(
         metric: Optional[str] = None,
         requested_indexing_policy: Optional[Dict[str, Any]] = None,
         default_indexing_policy: Optional[Dict[str, Any]] = None,
+        collection_vector_service_options: Optional[
+            CollectionVectorServiceOptions
+        ] = None,
     ) -> None:
         super().__init__(
             token, api_endpoint, astra_db_client, async_astra_db_client, namespace
@@ -126,12 +130,18 @@ async def _setup_db() -> None:
                 else:
                     dimension = embedding_dimension
 
+                # Used for enabling $vectorize on the collection
+                service_dict: Optional[Dict[str, Any]] = None
+                if collection_vector_service_options is not None:
+                    service_dict = collection_vector_service_options.as_dict()
+
                 try:
                     await async_astra_db.create_collection(
                         collection_name,
                         dimension=dimension,
                         metric=metric,
                         options=_options,
+                        service_dict=service_dict,
                     )
                 except (APIRequestError, ValueError):
                     # possibly the collection is preexisting and may have legacy,
@@ -161,12 +171,18 @@ async def _setup_db() -> None:
                     "set to False"
                 )
             else:
+                # Used for enabling $vectorize on the collection
+                service_dict: Optional[Dict[str, Any]] = None
+                if collection_vector_service_options is not None:
+                    service_dict = collection_vector_service_options.as_dict()
+
                 try:
                     self.astra_db.create_collection(
                         collection_name,
                         dimension=embedding_dimension,  # type: ignore[arg-type]
                         metric=metric,
                         options=_options,
+                        service_dict=service_dict,
                     )
                 except (APIRequestError, ValueError):
                     # possibly the collection is preexisting and may have legacy,