Struct SparseTermScorer 

pub struct SparseTermScorer<'a> { /* private fields */ }

Scorer for sparse vector dimensions

Wraps a BlockSparsePostingList with a query weight to implement ScoringIterator.

Implementations

impl<'a> SparseTermScorer<'a>

pub fn new(posting_list: &'a BlockSparsePostingList, query_weight: f32) -> Self

Create a new sparse term scorer

pub fn from_arc( posting_list: &'a Arc<BlockSparsePostingList>, query_weight: f32, ) -> Self

Create from Arc reference (for use with shared posting lists)

Trait Implementations

impl ScoringIterator for SparseTermScorer<'_>

fn doc(&self) -> DocId

Current document ID (u32::MAX if exhausted)

fn advance(&mut self) -> DocId

Advance to next document, returns new doc ID

fn seek(&mut self, target: DocId) -> DocId

Seek to first document >= target, returns new doc ID

fn score(&self) -> f32

Score contribution for current document

fn max_score(&self) -> f32

Maximum possible score for this term/dimension (global upper bound)

fn current_block_max_score(&self) -> f32

Current block’s maximum score upper bound (for block-level pruning)

fn is_exhausted(&self) -> bool

Check if iterator is exhausted

fn skip_to_next_block(&mut self) -> DocId

Skip to the next block, returning the first doc_id in the new block. Used for block-max WAND optimization when current block can’t beat threshold. Default implementation just advances (no block-level skipping).