Struct CachedEmbedder 

pub struct CachedEmbedder { /* private fields */ }

Caching wrapper around any Embedder.

Intercepts embed() calls and returns cached vectors for previously-seen query strings. All other trait methods delegate directly to the inner embedder.

Construction

use frankensearch_embed::CachedEmbedder;

let inner: Arc<dyn Embedder> = /* ... */;
let cached = CachedEmbedder::new(inner, 128);

Implementations

impl CachedEmbedder

pub fn new(inner: Arc<dyn Embedder>, capacity: usize) -> Self

Wrap an embedder with a bounded query cache.

capacity controls the maximum number of cached embeddings before FIFO eviction begins.

pub fn with_default_capacity(inner: Arc<dyn Embedder>) -> Self

Wrap an embedder with the default capacity (128 entries).

pub fn cache_stats(&self) -> CacheStats

Return a snapshot of cache statistics.

pub fn clear_cache(&self)

Clear all cached embeddings and reset statistics.

pub fn inner(&self) -> &dyn Embedder

Reference to the inner embedder.

Trait Implementations

impl Debug for CachedEmbedder

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Embedder for CachedEmbedder

fn embed<'a>(&'a self, cx: &'a Cx, text: &'a str) -> SearchFuture<'a, Vec<f32>>

Embed a single text string into a vector of f32 floats.

fn embed_batch<'a>( &'a self, cx: &'a Cx, texts: &'a [&'a str], ) -> SearchFuture<'a, Vec<Vec<f32>>>

Embed a batch of text strings.

fn dimension(&self) -> usize

The dimensionality of embedding vectors produced by this model.

fn id(&self) -> &str

A unique, stable identifier for this embedder.

fn model_name(&self) -> &str

Human-readable model name.

fn is_ready(&self) -> bool

Whether this embedder is loaded and operational.

fn is_semantic(&self) -> bool

Whether this embedder produces semantically meaningful vectors.

fn category(&self) -> ModelCategory

The speed/quality category of this embedder.

fn tier(&self) -> ModelTier

Default progressive tier assignment.

fn supports_mrl(&self) -> bool

Whether this model supports Matryoshka Representation Learning (dimension truncation for faster search with controlled quality loss).

fn truncate_embedding( &self, embedding: &[f32], target_dim: usize, ) -> Result<Vec<f32>, SearchError>

Truncate and re-normalize embedding to target_dim.