Struct Index 

pub struct Index { /* private fields */ }

In-memory flat ANN index.

Implementations

impl Index

pub fn new(metric: Metric) -> Self

Build an empty index.

pub fn metric(&self) -> Metric

Active metric.

pub fn len(&self) -> usize

Number of indexed vectors.

pub fn is_empty(&self) -> bool

True iff no vectors are indexed.

pub fn dim(&self) -> Option<usize>

Dimension of stored vectors, set on first insert.

pub fn add(&mut self, id: impl Into<String>, vector: &[f32]) -> Result<()>

Insert a single vector.

pub fn remove(&mut self, id: &str) -> bool

Remove the first entry whose id matches. Returns true if found. O(n) — uses swap-remove so the rest of the index isn’t shifted. dim is preserved even after the index becomes empty.

pub fn save<P: AsRef<Path>>(&self, path: P) -> Result<()>

Persist the entire index (metric, dim, ids, vectors) to a JSON file. Re-load with Index::load. JSON is verbose for f32 arrays but cross-platform and debuggable; binary persistence is on the v0.2 list.

pub fn load<P: AsRef<Path>>(path: P) -> Result<Self>

Reverse of Index::save. Loads an index previously saved with the same crate version.

pub fn add_batch( &mut self, ids: Vec<String>, matrix: &ArrayView2<'_, f32>, ) -> Result<()>

Insert many vectors. matrix is (n, d).

pub fn search(&self, query: &[f32], k: usize) -> Result<Vec<Hit>>

Top-k search. query is a 1-D vector of the same dimension as stored vectors.

pub fn search_batch( &self, queries: &ArrayView2<'_, f32>, k: usize, parallel: bool, ) -> Result<Vec<Vec<Hit>>>

Batch search. queries is (n_q, d). With parallel = true, each query runs on a rayon thread.

pub fn vectors(&self) -> Result<Array2<f32>>

Snapshot vectors as a (n, d) matrix (allocates a copy).

Trait Implementations

impl<'de> Deserialize<'de> for Index

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for Index

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.