rag-vector/ai.koog.rag.vector.storage/EmbeddingStorage

EmbeddingStorage

open class EmbeddingStorage<Document>(embedder: DocumentEmbedder<Document>, storage: VectorStorageBackend<Document>) : VectorStorage<Document, SimilaritySearchRequest> , CapabilityAwareStorage(source)

A VectorStorage naive implementation that composes a DocumentEmbedder with a VectorStorageBackend.

The embedding step is handled by this class, while the actual persistence is delegated to the provided VectorStorageBackend. This separation allows swapping backends (in-memory, file-based, real vector database) independently from the embedding model.

Subclasses can customize embedding, scoring, or request handling by overriding the protected template methods embedDocument, embedQuery, and score, without rewriting entire storage operations.

Type Parameters

Document

The type of the document being stored and ranked.

Inheritors

FileDocumentEmbeddingStorage

InMemoryDocumentEmbeddingStorage

Constructors

EmbeddingStorage

constructor(embedder: DocumentEmbedder<Document>, storage: VectorStorageBackend<Document>)

Properties

capabilities

open override val capabilities: Set<StorageCapability>

Functions

add

open suspend override fun add(documents: List<Document>, namespace: String?): List<String>

allDocuments

open fun allDocuments(): Flow<Document>

Retrieves a flow of all documents stored in the system.

delete

open suspend override fun delete(ids: List<String>, namespace: String?): List<String>

get

open suspend override fun get(ids: List<String>, namespace: String?): List<Document>

rankDocuments

open override fun ~~rankDocuments~~(query: String): Flow<SearchResult<Document>>

open suspend override fun search(request: SimilaritySearchRequest, namespace: String?): List<SearchResult<Document>>

supports

open fun supports(capability: StorageCapability): Boolean

update

open suspend override fun update(documents: Map<String, Document>, namespace: String?): List<String>