Enum IndexKind

Source

#[non_exhaustive]pub enum IndexKind {
    Range,
    Unique,
    Text,
    Ann,
    Agg,
}

Expand description

Index kind (KIND range|unique).

use kevy_index::IndexKind;
assert_eq!(IndexKind::parse(b"UNIQUE"), Some(IndexKind::Unique));
assert_eq!(IndexKind::parse(IndexKind::Agg.tag().as_bytes()), Some(IndexKind::Agg));
assert_eq!(IndexKind::parse(b"hash"), None);

Variants (Non-exhaustive)§

This enum is marked as non-exhaustive

Non-exhaustive enums could have additional variants added in future. Therefore, when matching against variants of non-exhaustive enums, an extra wildcard arm must be added to account for any future variants.

§

Range

Ordered scan over (value, key) pairs.

use kevy_index::{IndexValue, Segment};
let mut seg = Segment::new();
seg.apply(b"u:1", None, Some(IndexValue::I64(30)));
seg.apply(b"u:2", None, Some(IndexValue::I64(20)));
let (hits, _) = seg.range(&IndexValue::I64(0), &IndexValue::I64(99), None, 10);
assert_eq!(hits[0].0, b"u:2", "value order, not key order");

§

Unique

Point lookup by value; duplicates recorded (declarative fence: uniqueness is verified, not write-enforced).

use kevy_index::{IndexValue, Segment};
let mut seg = Segment::new();
seg.apply(b"u:1", None, Some(IndexValue::Str(b"a@x".to_vec())));
seg.apply(b"u:2", None, Some(IndexValue::Str(b"a@x".to_vec())));
// the second write is not refused; the lookup shows the duplicate
assert_eq!(seg.eq(&IndexValue::Str(b"a@x".to_vec()), 10).len(), 2);

§

Text

Full-text: the field tokenizes into an inverted segment (kevy-text); queried with MATCH, BM25-ranked.

use kevy_index::{FieldSpec, IndexKind, IndexSpec, ValType};
let fields = vec![FieldSpec::new("title").with_weight(2.0), FieldSpec::new("body")];
let text = IndexSpec::builder("posts", "post:", IndexKind::Text, ValType::Str)
    .with_fields(fields)
    .build()?;
assert_eq!(text.fields().len(), 2, "only text serves several fields");

§

Ann

ANN: the field holds an f32 LE vector indexed in an HNSW graph (kevy-vector); queried with KNN, distance-ranked.

use kevy_index::{AnnSpec, IndexKind, IndexSpec, ValType};
let b = IndexSpec::builder("emb", "doc:", IndexKind::Ann, ValType::Vector).with_field("v");
assert!(b.clone().build().is_err(), "an ANN index needs its parameters");
assert_eq!(b.with_ann(AnnSpec::new(128)).build()?.ann(), Some(AnnSpec::new(128)));

§

Agg

Aggregate: per-group count/sum/min/max of the field, grouped by IndexSpec::group_by; queried with GROUP/GROUPS.

use kevy_index::{IndexKind, IndexSpec, ValType};
let spend = IndexSpec::builder("spend", "o:", IndexKind::Agg, ValType::F64)
    .with_field("total")
    .with_group_by("region")
    .build()?;
assert_eq!(spend.group_by(), Some(&b"region"[..]));