Struct TextIndex 

pub struct TextIndex { /* private fields */ }

In-memory BM25 postings index for one node (label, property) pair.

Implementations

impl TextIndex

pub fn search_candidates( &self, query: &str, candidates: &[NodeId], k: usize, ) -> Vec<TextSearchHit>

Rank explicit node candidates for query using this index’s BM25 corpus stats.

Candidate ids are deduplicated and missing/non-indexed ids are ignored. Term document frequencies and average document length remain global to the maintained index, so this returns the same ordering as a full Self::search followed by candidate filtering when enough global hits are materialized.

pub fn search_candidates_checked( &self, query: &str, candidates: &[NodeId], k: usize, checker: CancellationChecker<'_>, ) -> Result<Vec<TextSearchHit>, TextSearchError>

Rank explicit node candidates for query with cooperative cancellation checks.

Errors

Returns TextSearchError::Cancelled, TextSearchError::Timeout, or TextSearchError::NodeScanBudgetExceeded when the supplied checker trips while deduplicating or scoring candidates.

impl TextIndex

pub fn build( graph: &SeleneGraph, label: DbString, property: DbString, ) -> GraphResult<Self>

Build a postings index from the current graph snapshot.

Only alive nodes that carry label and whose property value is a non-empty string document are indexed. Non-string values are ignored, which matches the exact BM25 scan path.

Errors

Returns GraphError::Inconsistent if the label index references a row without a resolvable node id or property row.

pub fn empty(label: DbString, property: DbString) -> Self

Construct an empty postings index for label.property.

pub const fn label(&self) -> &DbString

Return the indexed node label.

pub const fn property(&self) -> &DbString

Return the indexed node property.

pub const fn rows(&self) -> &RoaringBitmap

Return the indexed row bitmap.

pub fn document_count(&self) -> usize

Return the number of indexed string documents.

pub fn term_count(&self) -> usize

Return the number of distinct indexed terms.

pub const fn posting_count(&self) -> usize

Return the total number of term-document postings.

pub fn stats(&self) -> TextIndexStats

Return aggregate index counters.

pub fn memory_usage(&self) -> TextIndexMemoryUsage

Return an estimated memory usage snapshot for this index.

pub fn search(&self, query: &str, k: usize) -> Vec<TextSearchHit>

Rank indexed documents for query using BM25.

pub fn search_checked( &self, query: &str, k: usize, checker: CancellationChecker<'_>, ) -> Result<Vec<TextSearchHit>, TextSearchError>

Rank indexed documents for query with cooperative cancellation checks.

The query path visits only postings for query terms, not every indexed document. Scores and tie ordering match the exact scan oracle.

Errors

Returns TextSearchError::Cancelled, TextSearchError::Timeout, or TextSearchError::NodeScanBudgetExceeded when the supplied checker trips while collecting postings or scoring candidate documents.

Trait Implementations

impl Clone for TextIndex

fn clone(&self) -> TextIndex

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for TextIndex

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

Formats the value using the given formatter.