Struct RowUniquenessTracker 

pub struct RowUniquenessTracker { /* private fields */ }

Full-stream row-duplicate tracking with bounded memory.

Every record’s fields are folded into a RowSignature and fed to a CardinalityEstimator: duplicates are counted exactly while the distinct-row signatures fit the estimator’s exact set, and estimated (flagged approximate) once it spills to its HLL sketch. Unlike the per-column reservoirs, this sees whole rows — including null-like values — so the count is row-aligned by construction.

Implementations

impl RowUniquenessTracker

pub fn observe(&mut self, signature: String)

pub fn rows_seen(&self) -> usize

pub fn duplicate_rows(&self) -> usize

Rows minus distinct rows; exact until the estimator spills.

pub fn is_approximate(&self) -> bool

pub fn mark_incomparable(&mut self)

Record that the signatures folded in so far were not all made against the same schema, so they cannot be compared with each other.

pub fn merge(&mut self, other: &RowUniquenessTracker)

pub fn memory_usage_bytes(&self) -> usize

pub fn exact_bytes(&self) -> usize

Heap held by the exact set of row signatures: what Self::spill frees.

pub fn spill(&mut self)

Answer the duplicate count from the sketch from here on, freeing the exact set. For memory pressure; the count is then reported approximate.

pub fn summary(&self) -> Option<RowDuplicateSummary>

Summary for quality metrics, or None when no rows were observed (e.g. an engine that never fed whole records through this tracker) or when the signatures seen are not comparable with one another.

Trait Implementations

impl Clone for RowUniquenessTracker

fn clone(&self) -> Self

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for RowUniquenessTracker

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

Formats the value using the given formatter.

impl Default for RowUniquenessTracker

fn default() -> Self

Returns the “default value” for a type.