Struct SparseBucket 

pub struct SparseBucket {
    pub entries: Vec<SparseEntry>,
    pub extra: BTreeMap<String, Ipld>,
}

Per-node bucket of sparse embeddings inside the Commit.sparse Prolly tree.

One bucket per node. Each bucket holds a sorted (vocab_id, SparseEmbed) list so a node may carry multiple sparse embeddings simultaneously - e.g. one BGE-M3 vector plus one OpenSearch-distill vector for the same chunk text. Lookups index into the bucket by vocab_id string after the outer Prolly walk has returned the bucket itself.

Fields

entries: Vec<SparseEntry>

Entries sorted lexicographically by vocab_id for byte-stable canonical form. The sort is enforced on every serialize, so callers may push entries in any order without breaking CID determinism on the bucket itself.

extra: BTreeMap<String, Ipld>

Forward-compat extension carrier. Unknown CBOR fields land here on decode and are emitted verbatim on re-encode, so a future schema bump that adds a per-bucket field stays round-trippable on today’s reader.

Implementations

impl SparseBucket

pub const KIND: &'static str = "sparse_bucket"

On-wire _kind discriminator. Every content-addressed object in mnem/0.x carries a _kind field as the first canonical key so a corrupt bucket / wrong-type decode fails fast with an actionable error instead of silently mis-decoding.

pub fn get(&self, vocab_id: &str) -> Option<&SparseEmbed>

Look up a sparse embedding by vocab_id string. Returns None when this bucket has no entry for the requested vocabulary; the caller typically falls back to lazy compute via the configured sparse encoder adapter.

pub fn upsert(&mut self, vocab_id: String, sparse: SparseEmbed)

Insert or replace an entry by vocab_id. The bucket does not return the previous value (unlike EmbeddingBucket::upsert) because SparseEmbed contains Vec<f32> which is not PartialEq in a meaningful sense for callers here.

pub fn remove(&mut self, vocab_id: &str)

Remove an entry by vocab_id.

Trait Implementations

impl Clone for SparseBucket

fn clone(&self) -> SparseBucket

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for SparseBucket

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

Formats the value using the given formatter.

impl Default for SparseBucket

fn default() -> SparseBucket

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for SparseBucket

fn deserialize<D: Deserializer<'de>>(deserializer: D) -> Result<Self, D::Error>

Deserialize this value from the given Serde deserializer.

impl Serialize for SparseBucket

fn serialize<S: Serializer>(&self, serializer: S) -> Result<S::Ok, S::Error>

Serialize this value into the given Serde serializer.