Struct SemanticCache 

pub struct SemanticCache {
    pub config: CacheConfig,
    pub distance_threshold: f32,
    pub vector_dimensions: usize,
    pub dtype: VectorDataType,
    pub index: SearchIndex,
    /* private fields */
}

Semantic cache backed by a Redis Search vector index.

Fields

config: CacheConfig

Cache configuration.

distance_threshold: f32

Distance threshold used for semantic hits.

vector_dimensions: usize

Prompt embedding dimensions stored in Redis.

dtype: VectorDataType

Vector element data type used for the index schema.

index: SearchIndex

Underlying search index.

Implementations

impl SemanticCache

pub fn new( config: CacheConfig, distance_threshold: f32, vector_dimensions: usize, ) -> Result<Self>

Creates a new semantic cache with the default reserved schema.

pub fn with_dtype( config: CacheConfig, distance_threshold: f32, vector_dimensions: usize, dtype: VectorDataType, ) -> Result<Self>

Creates a new semantic cache with a specific vector data type.

pub fn with_filterable_fields( config: CacheConfig, distance_threshold: f32, vector_dimensions: usize, filterable_fields: &[Value], ) -> Result<Self>

Creates a new semantic cache with additional filterable schema fields.

pub fn with_options( config: CacheConfig, distance_threshold: f32, vector_dimensions: usize, dtype: VectorDataType, filterable_fields: &[Value], ) -> Result<Self>

Creates a new semantic cache with full control over dtype and filterable fields.

pub fn with_vectorizer<V>(self, vectorizer: V) -> Self

where V: Vectorizer + 'static,

Attaches a synchronous vectorizer used when callers pass prompts instead of vectors.

pub fn with_default_vectorizer(self) -> Result<Self>

Attaches the default HuggingFace local vectorizer.

This uses the AllMiniLML6V2 model from fastembed which runs locally via ONNX Runtime and requires no API key. The model is downloaded from HuggingFace Hub on first use.

Errors

Returns an error if the model cannot be loaded.

pub fn set_vectorizer<V>(&mut self, vectorizer: V)

where V: Vectorizer + 'static,

Replaces the vectorizer used for prompt embedding.

pub fn ttl(&self) -> Option<u64>

Returns the configured default TTL for semantic cache entries.

pub fn set_ttl(&mut self, ttl_seconds: Option<u64>)

Sets or clears the default TTL for semantic cache entries.

pub fn set_threshold(&mut self, distance_threshold: f32) -> Result<()>

Updates the semantic distance threshold.

pub fn store( &self, prompt: &str, response: &str, vector: Option<&[f32]>, metadata: Option<Value>, filters: Option<Map<String, Value>>, ttl_seconds: Option<u64>, ) -> Result<String>

Stores a semantic cache entry and returns its Redis key.

pub async fn astore( &self, prompt: &str, response: &str, vector: Option<&[f32]>, metadata: Option<Value>, filters: Option<Map<String, Value>>, ttl_seconds: Option<u64>, ) -> Result<String>

Stores a semantic cache entry asynchronously and returns its Redis key.

pub fn check( &self, prompt: Option<&str>, vector: Option<&[f32]>, num_results: usize, return_fields: Option<&[&str]>, filter_expression: Option<FilterExpression>, distance_threshold: Option<f32>, ) -> Result<Vec<Map<String, Value>>>

Checks the semantic cache for similar prompts or a supplied vector.

pub async fn acheck( &self, prompt: Option<&str>, vector: Option<&[f32]>, num_results: usize, return_fields: Option<&[&str]>, filter_expression: Option<FilterExpression>, distance_threshold: Option<f32>, ) -> Result<Vec<Map<String, Value>>>

Checks the semantic cache for similar prompts or a supplied vector asynchronously.

pub fn update(&self, key: &str, fields: Map<String, Value>) -> Result<()>

Updates cached fields for a stored semantic cache entry and refreshes TTL.

pub async fn aupdate(&self, key: &str, fields: Map<String, Value>) -> Result<()>

Updates cached fields asynchronously for a stored semantic cache entry and refreshes TTL.

pub fn clear(&self) -> Result<usize>

Clears all semantic cache entries while preserving the index.

pub async fn aclear(&self) -> Result<usize>

Clears all semantic cache entries asynchronously while preserving the index.

pub fn delete(&self) -> Result<()>

Deletes the semantic cache index and its documents.

pub async fn adelete(&self) -> Result<()>

Deletes the semantic cache index asynchronously and its documents.

pub fn drop_ids(&self, ids: &[String]) -> Result<()>

Drops stored entries by their entry ids.

pub fn drop_keys(&self, keys: &[String]) -> Result<()>

Drops stored entries by their Redis keys.

pub async fn adrop_ids(&self, ids: &[String]) -> Result<()>

Drops stored entries asynchronously by their entry ids.

pub async fn adrop_keys(&self, keys: &[String]) -> Result<()>

Drops stored entries asynchronously by their Redis keys.

Trait Implementations

impl Clone for SemanticCache

fn clone(&self) -> SemanticCache

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for SemanticCache

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

Formats the value using the given formatter.