Sekejap DB

A graph-first, embedded multi-model database engine for Rust and Python.

Graph is the primary structure. Vector, Spatial, and Full-Text are first-class attributes on nodes, queryable in the same pipeline. Runs in-process — zero network overhead.

Features

• Graph-First: Edges are first-class. Queries traverse the graph to prune the search space before applying other indexes.
• HNSW Engine: Parallel construction (Rayon), NEON/AVX2 SIMD CosineDistance, ef_construction=32. 44x faster batch build vs v0.2.
• Spatial (PostGIS-par): R-Tree index on full GeoJSON geometry envelopes. Point, Polygon, LineString, Multi* — with DE-9IM predicates (st_within, st_contains, st_intersects, st_dwithin).
• Full-Text: Pluggable adapters — Tantivy (default) or SeekStorm.
• Embedded: Memory-mapped arenas, zero-copy storage.
• SekejapQL: Lightweight text query language — one op per line, compiles to the same pipeline as JSON.
• CLI: Interactive REPL (skcli) — SekejapQL and JSON queries, mutations, introspection.

Quick Start

Python

import sekejap, json

db = sekejap.SekejapDB("./data", capacity=1_000_000)
db.init_hnsw(16)
db.init_fulltext()

# Write
db.put("persons/lucci", json.dumps({
    "name": "Rob Lucci", "status": "wanted",
    "vectors": {"dense": [0.12, 0.87, ...]},
    "geo": {"loc": {"lat": 3.1105, "lon": 101.6682}}
}))
db.link("persons/lucci", "crimes/robbery-001", "committed", 1.0)

# Query — SekejapQL
result = db.query_skql("""
    one "crimes/robbery-001"
    backward "committed"
    where_eq "status" "wanted"
    forward "lives_at"
    near 3.1291 101.6710 5.0
    select "name" "alias" "geo"
""")

db.flush()
db.close()

Rust

use sekejap::SekejapDB;

let db = SekejapDB::new(std::path::Path::new("./data"), 1_000_000)?;
db.init_hnsw(16);
db.init_fulltext(std::path::Path::new("./data"));

db.nodes().put("persons/lucci", r#"{"name":"Rob Lucci","status":"wanted"}"#)?;
db.edges().link("persons/lucci", "crimes/robbery-001", "committed", 1.0)?;

// SekejapQL (auto-detected — anything not starting with '{')
let result = db.query("collection \"crimes\"\nwhere_eq \"type\" \"robbery\"\ntake 20")?;

// JSON pipeline (auto-detected — starts with '{')
let result = db.query(r#"{"pipeline":[{"op":"collection","name":"crimes"},{"op":"take","n":5}]}"#)?;

// Fluent builder (Rust only)
let result = db.nodes().collection("crimes")
    .where_eq("type", serde_json::json!("robbery"))
    .near(3.13, 101.67, 1.0)
    .take(20)
    .collect()?;

SekejapQL — One Op Per Line

# Find all gang members → their crimes → near a location
one "persons/lucci"
forward "member_of"
backward "member_of"
forward "committed"
near 3.1291 101.6710 2.0
sort "severity" desc
take 10
select "type" "severity" "occurred_at"

# News → crime → suspect network inside a polygon
collection "articles"
matching "armed robbery Bangsar 2024"
forward "reported_by"
backward "committed"
forward "member_of"
backward "member_of"
spatial_within_polygon (3.128,101.665) (3.135,101.665) (3.135,101.678) (3.128,101.678) (3.128,101.665)
where_eq "status" "wanted"
select "name" "alias" "geo" "priors"

# PostGIS-par geometry predicates
collection "zones"
st_intersects (3.128,101.665) (3.128,101.678) (3.135,101.678) (3.135,101.665) (3.128,101.665)

Pipe style also works: collection "crimes" | where_eq "type" "robbery" | near 3.1 101.6 1.0 | take 20

CLI (skcli)

cargo run -p skcli -- --path ./data

sekejap> collection "crimes" | where_eq "type" "robbery" | take 10
sekejap> one "persons/lucci" | forward "committed"
sekejap> count collection "crimes"
sekejap> explain collection "crimes" | take 5
sekejap> mutate {"mutation":"put_json","data":{"_id":"crimes/001","type":"robbery"}}
sekejap> \d crimes
sekejap> \l

Benchmark (10k Records, Apple Silicon)

See docs/benchmark-results.md for full Rust vs Python results and methodology.

Full Documentation

See the docs/ folder:

• docs/api-reference.md — Complete API reference (setup, schema, mutations, all SekejapQL ops, JSON pipeline, data conventions, performance)
• docs/interface-table.md — Interface inventory and status
• docs/roadmap.md — Development roadmap and execution notes
• docs/benchmark-results.md — Benchmark results (Rust vs Python vs SQLite)

Build

cargo test
cargo build --release

# CLI
cargo run -p skcli -- --path ./data

# Python wheel
maturin develop --release

License

MIT