Struct SemanticCache 

pub struct SemanticCache { /* private fields */ }

A cache that matches queries by embedding similarity rather than exact string match.

Implementations

impl SemanticCache

pub fn new(config: SemanticCacheConfig) -> Self

Create a new cache with the supplied configuration.

pub fn get(&self, query_embedding: &[f32]) -> Option<Value>

Try to find a cached result for query_embedding.

Performs a linear scan over all entries comparing cosine similarity. Returns the result of the best-matching non-expired entry whose similarity is at or above the configured threshold.

pub fn put(&self, query_embedding: Vec<f32>, query_text: String, results: Value)

Store results under query_embedding.

If the cache is at capacity, the oldest entry (by created_at) is evicted first.

pub fn invalidate_memory(&self, memory_id: i64)

Remove all entries whose result JSON contains memory_id.

The JSON is checked for any occurrence of the integer value at any position in the document.

pub fn clear(&self)

Remove all cached entries.

pub fn stats(&self) -> SemanticCacheStats

Return a snapshot of cache statistics.