SQLiteGraph

Embedded graph database with dual backend architecture.

Positioning: Single-binary embedded database (no server). Persistent storage with atomic batch commits. Graph algorithms + HNSW vector search in one engine. SQLite backend: best for point lookups. Native V3: 10-20× faster for traversals.

What's New in v2.0.0

Native V3 Backend - Production-ready high-performance backend

• B+Tree-based storage with unlimited node capacity (no 2048 limit)
• 10-20× faster traversals compared to SQLite backend
• Full feature parity: graph algorithms, HNSW vectors, Pub/Sub, KV store
• WAL integration for durability and crash recovery

SQLite Backend Pub/Sub - Now fully supported

• In-process event notification for graph changes
• Multiple subscriber support with filtered events
• Works across all backends

HNSW Vector Storage for V3 - Vector search with V3 backend

• V3VectorStorage - stores vectors in V3's KV store
• Same HNSW algorithm, different persistence layer
• Unified API across backends

Reshaped CLI (sqlitegraph-cli crate)

• Read-only by default (--write flag for modifications)
• Cypher-like query support: MATCH (n:User) RETURN n.name
• Dual backend support (--backend sqlite|v3)

Backends

Quick Start

[dependencies]
# SQLite backend (default)
sqlitegraph = "2.0"

# OR Native V3 backend (faster traversals)
sqlitegraph = { version = "2.0", features = ["native-v3"] }

use sqlitegraph::backend::{GraphBackend, NodeSpec};
use sqlitegraph::backend::sqlite::SqliteGraphBackend;

fn main() -> Result<(), Box<dyn std::error::Error>> {
    let backend = SqliteGraphBackend::in_memory()?;
    
    let node_id = backend.insert_node(NodeSpec {
        kind: "User".to_string(),
        name: "Alice".to_string(),
        file_path: None,
        data: serde_json::json!({"age": 30}),
    })?;
    
    println!("Created node: {}", node_id);
    Ok(())
}

Features

• Dual Backend Architecture: SQLite (stable, debuggable) or Native V3 (high performance)
• Graph Algorithms: 35+ algorithms including BFS, DFS, PageRank, community detection
• HNSW Vector Search: Approximate nearest neighbor search with multiple distance metrics
• ACID Transactions: Full durability with WAL support
• Pub/Sub Events: Real-time notifications for graph changes
• KV Store: Key-value storage with TTL and MVCC

Ecosystem

Documentation

• API Documentation
• Architecture Guide
• Graph Algorithms
• Changelog

License

GPL-3.0-or-later