Struct IndexMerger 

pub struct IndexMerger { /* private fields */ }

Combines multiple FlatIndex instances into a single merged index.

Deduplication

When two entries share the same id, the last one wins (insertion order across indices, then within each index).

Implementations

impl IndexMerger

pub fn new() -> Self

Create a new, empty merger.

pub fn add_index(&mut self, idx: FlatIndex)

Add an index to the merge set.

pub fn merge(&mut self) -> Result<FlatIndex, MergeError>

Merge all added indices into a single FlatIndex.

Deduplication is performed on id: if multiple entries share the same ID, the one that appears latest (last index, last position within that index) wins.

Returns MergeError::EmptyInput if no indices have been added.

pub fn merge_with_filter<F>( &mut self, filter: F, ) -> Result<FlatIndex, MergeError>

where F: Fn(&VectorEntry) -> bool,

Merge all indices, retaining only entries for which filter returns true. Deduplication happens before filtering.

pub fn merge_with_stats( &mut self, ) -> Result<(FlatIndex, MergeStats), MergeError>

Merge all indices and return both the merged index and statistics.

pub fn split(idx: &FlatIndex, parts: usize) -> Vec<FlatIndex>

Split a FlatIndex into parts evenly-sized sub-indices.

If the number of entries does not divide evenly, the first entries.len() % parts partitions will each receive one extra entry.

Returns MergeError::InvalidParts if parts == 0.

Trait Implementations

impl Debug for IndexMerger

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

Formats the value using the given formatter.

impl Default for IndexMerger

fn default() -> IndexMerger

Returns the “default value” for a type.