Struct SemanticQueryCache 

pub struct SemanticQueryCache { /* private fields */ }

An in-memory cache for semantic search results with:

• FNV-1a fingerprinting — O(d) key computation over the query embedding.
• TTL expiry — stale entries are rejected on lookup and purged by Self::evict_expired.
• LRU eviction — when the cache is full the least-recently-used entry (front of access_order) is discarded.
• Hit/miss statistics — exposed through Self::stats.

Implementations

impl SemanticQueryCache

pub fn new(config: QueryCacheConfig) -> Self

Create a new, empty cache with the given QueryCacheConfig.

pub fn get(&mut self, embedding: &[f32], current_tick: u64) -> Option<Vec<u64>>

Look up cached results for embedding at the given current_tick.

Return value

• Some(doc_ids) — cache hit; doc_ids is a clone of the stored result vector.
• None — cache miss (unknown key or expired entry).

pub fn put( &mut self, embedding: &[f32], result_doc_ids: Vec<u64>, current_tick: u64, ttl_override: Option<u64>, )

Store result_doc_ids for embedding in the cache.

If an entry already exists for this embedding it is refreshed in-place (TTL reset, results replaced, hit_count zeroed). Otherwise a new entry is created, potentially evicting the LRU entry when the cache is full.

Pass ttl_override to use a TTL other than QueryCacheConfig::default_ttl_ticks for this specific entry.

pub fn invalidate(&mut self, embedding: &[f32]) -> bool

Remove the cache entry for embedding.

Returns true if an entry existed and was removed, false otherwise.

pub fn evict_expired(&mut self, current_tick: u64) -> usize

Remove all entries whose TTL has elapsed at current_tick.

Returns the number of entries removed. Each removed entry increments QueryCacheStats::expirations.

pub fn stats(&self) -> &QueryCacheStats

Return a shared reference to the accumulated cache statistics.