Struct ChunkedVectorStore 

pub struct ChunkedVectorStore { /* private fields */ }

Chunked Vector Store

Implementations

impl ChunkedVectorStore

pub fn new( document_store: Arc<InMemoryChunkedDocumentStore>, vector_size: usize, ) -> ChunkedVectorStore

Creates a new ChunkedVectorStore

pub async fn add_chunk_vector( &self, chunk_id: impl Into<String>, embedding: Vec<f32>, ) -> Result<(), VectorStoreError>

Adds a chunk vector (chunk_id + embedding)

pub async fn add_chunk_vectors( &self, chunk_ids: Vec<String>, embeddings: Vec<Vec<f32>>, ) -> Result<(), VectorStoreError>

Adds chunk vectors in bulk

pub async fn add_parent_document( &self, document: Document, chunk_size: usize, embeddings_fn: impl Fn(&str) -> Vec<f32>, ) -> Result<(String, Vec<String>), VectorStoreError>

Adds from a Parent document (auto-split + vectorized)

pub async fn get_embedding( &self, chunk_id: &str, ) -> Result<Option<Vec<f32>>, VectorStoreError>

Gets the vector for a chunk_id (M4: O(1) HashMap lookup)

pub async fn vector_count(&self) -> usize

Gets the number of stored vectors

Trait Implementations

impl VectorStore for ChunkedVectorStore

fn similarity_search_with_filter<'life0, 'life1, 'life2, 'async_trait>( &'life0 self, query_embedding: &'life1 [f32], k: usize, filter: Option<&'life2 MetadataFilter>, ) -> Pin<Box<dyn Future<Output = Result<Vec<SearchResult>, VectorStoreError>> + Send + 'async_trait>>

where 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait, ChunkedVectorStore: 'async_trait,

S3: chunked-store metadata filtering.

The vector index carries no metadata, so filtering requires fetching documents from the document store by chunk_id. The semantics are “score everything → scan descending, filtering each item by metadata → stop once k items are collected”, guaranteeing the top-k by similarity after filtering (rather than truncating first and filtering later).

fn add_documents<'life0, 'async_trait>( &'life0 self, documents: Vec<Document>, embeddings: Vec<Vec<f32>>, ) -> Pin<Box<dyn Future<Output = Result<Vec<String>, VectorStoreError>> + Send + 'async_trait>>

where 'life0: 'async_trait, ChunkedVectorStore: 'async_trait,

Adds documents.

fn similarity_search<'life0, 'life1, 'async_trait>( &'life0 self, query_embedding: &'life1 [f32], k: usize, ) -> Pin<Box<dyn Future<Output = Result<Vec<SearchResult>, VectorStoreError>> + Send + 'async_trait>>

where 'life0: 'async_trait, 'life1: 'async_trait, ChunkedVectorStore: 'async_trait,

Searches similar documents.

fn similarity_search_with_min_score<'life0, 'life1, 'async_trait>( &'life0 self, query_embedding: &'life1 [f32], k: usize, min_score: Option<f32>, ) -> Pin<Box<dyn Future<Output = Result<Vec<SearchResult>, VectorStoreError>> + Send + 'async_trait>>

where 'life0: 'async_trait, 'life1: 'async_trait, ChunkedVectorStore: 'async_trait,

Similarity search with a minimum score threshold.

fn get_document<'life0, 'life1, 'async_trait>( &'life0 self, id: &'life1 str, ) -> Pin<Box<dyn Future<Output = Result<Option<Document>, VectorStoreError>> + Send + 'async_trait>>

where 'life0: 'async_trait, 'life1: 'async_trait, ChunkedVectorStore: 'async_trait,

Gets document by ID.

fn get_embedding<'life0, 'life1, 'async_trait>( &'life0 self, id: &'life1 str, ) -> Pin<Box<dyn Future<Output = Result<Option<Vec<f32>>, VectorStoreError>> + Send + 'async_trait>>

where 'life0: 'async_trait, 'life1: 'async_trait, ChunkedVectorStore: 'async_trait,

Gets document embedding by ID.

fn delete_document<'life0, 'life1, 'async_trait>( &'life0 self, id: &'life1 str, ) -> Pin<Box<dyn Future<Output = Result<(), VectorStoreError>> + Send + 'async_trait>>

where 'life0: 'async_trait, 'life1: 'async_trait, ChunkedVectorStore: 'async_trait,

Deletes document.

fn count<'life0, 'async_trait>( &'life0 self, ) -> Pin<Box<dyn Future<Output = usize> + Send + 'async_trait>>

where 'life0: 'async_trait, ChunkedVectorStore: 'async_trait,

Returns document count.

fn clear<'life0, 'async_trait>( &'life0 self, ) -> Pin<Box<dyn Future<Output = Result<(), VectorStoreError>> + Send + 'async_trait>>

where 'life0: 'async_trait, ChunkedVectorStore: 'async_trait,

Clears store.

fn embed_query(&self) -> Option<&dyn Embeddings>

Returns this vector store’s built-in text embedder, if any.

fn similarity_search_text<'life0, 'life1, 'async_trait>( &'life0 self, query: &'life1 str, k: usize, ) -> Pin<Box<dyn Future<Output = Result<Vec<SearchResult>, VectorStoreError>> + Send + 'async_trait>>

where 'life0: 'async_trait, 'life1: 'async_trait, Self: 'async_trait,

Text similarity search: vectorizes query with the embedder returned by embed_query, then searches.