Trait VectorStore 

pub trait VectorStore: Send + Sync {
    // Required methods
    fn add_documents<'a>(
        &'a self,
        documents: &'a [Document],
        embeddings: &'a dyn Embeddings,
    ) -> BoxFuture<'a, Result<Vec<String>, SynwireError>>;
    fn similarity_search<'a>(
        &'a self,
        query: &'a str,
        k: usize,
        embeddings: &'a dyn Embeddings,
    ) -> BoxFuture<'a, Result<Vec<Document>, SynwireError>>;
    fn similarity_search_with_score<'a>(
        &'a self,
        query: &'a str,
        k: usize,
        embeddings: &'a dyn Embeddings,
    ) -> BoxFuture<'a, Result<Vec<(Document, f32)>, SynwireError>>;
    fn delete<'a>(
        &'a self,
        ids: &'a [String],
    ) -> BoxFuture<'a, Result<(), SynwireError>>;
}

Trait for vector stores that persist and query document embeddings.

Implementations manage storage of documents alongside their embedding vectors and support similarity-based retrieval.

Cancel safety

add_documents is not cancel-safe: dropping the future mid-write may leave a partial set of documents persisted. similarity_search and similarity_search_with_score are cancel-safe for read-only stores, but dropping mid-query may waste compute. delete is not cancel-safe: partial deletions are possible.

Required Methods

fn add_documents<'a>( &'a self, documents: &'a [Document], embeddings: &'a dyn Embeddings, ) -> BoxFuture<'a, Result<Vec<String>, SynwireError>>

Add documents to the store, returning their assigned IDs.

fn similarity_search<'a>( &'a self, query: &'a str, k: usize, embeddings: &'a dyn Embeddings, ) -> BoxFuture<'a, Result<Vec<Document>, SynwireError>>

Retrieve the k most similar documents to the query.

fn similarity_search_with_score<'a>( &'a self, query: &'a str, k: usize, embeddings: &'a dyn Embeddings, ) -> BoxFuture<'a, Result<Vec<(Document, f32)>, SynwireError>>

Retrieve the k most similar documents with their similarity scores.

fn delete<'a>( &'a self, ids: &'a [String], ) -> BoxFuture<'a, Result<(), SynwireError>>

Delete documents by their IDs.

Implementors

impl VectorStore for InMemoryVectorStore