Struct IndexCache 

pub struct IndexCache { /* private fields */ }

Lockfree-ish process cache of loaded indexes.

Implementations

impl IndexCache

pub fn new(model: StaticModel) -> Self

Build a cache with the default capacity (DEFAULT_CACHE_SIZE).

pub fn with_capacity(model: StaticModel, capacity: usize) -> Self

Build a cache with an explicit capacity.

pub async fn get_or_load( &self, repo: &str, include_text_files: bool, ) -> Result<Arc<RwLock<VelesIndex>>>

Get or lazily build the VelesIndex for repo.

Returns an Arc<RwLock<VelesIndex>> the caller can .read() or .write() independently of the cache lock. Multiple concurrent loaders of the same repo share a single in-flight build via the internal OnceCell.

repo is either a local directory path or an https:// git URL. Local paths prefer the persisted .veles/ index when one exists (fast load) and fall back to a fresh in-memory build otherwise.

pub fn peek(&self, repo: &str) -> Option<Arc<RwLock<VelesIndex>>>

Look up repo without building. Returns Some only if the cell has been initialised (i.e. a previous get_or_load for this repo has completed successfully). Used by callers that want to gate on “is this repo bootstrapped yet?” without triggering an expensive build — e.g. the gRPC GetStats RPC.

pub fn invalidate(&self, repo: &str) -> bool

Drop the cached entry for repo if present. Useful for tests and explicit invalidation.

pub fn len(&self) -> usize

Current number of cached repos.

pub fn is_empty(&self) -> bool

True if no repos are cached.

pub fn capacity(&self) -> usize

Configured capacity.