🧠 local_hnsw #

A lightweight, in-memory HNSW (Hierarchical Navigable Small World) vector index for Dart & Flutter. Supports fast approximate nearest neighbor (ANN) search with customizable distance metrics.

✨ Features #

• In-memory approximate nearest neighbor search
• Supports cosine and euclidean similarity
• Generic and type-safe (LocalHNSW<T>)
• Add, search, delete, save, and load vectors
• No native dependencies — pure Dart!

🚀 Getting Started #

1. Add dependency #

dependencies:
  local_hnsw: ^1.0.0

2. Import the package #

import 'package:local_hnsw/local_hnsw.dart';

✅ Example #

final index = LocalHNSW<String>(
  dim: 3,
  metric: LocalHnswMetric.cosine, // or .euclidean
);

index.add(LocalHnswItem(item: 'apple', vector: [0.1, 0.2, 0.3]));
index.add(LocalHnswItem(item: 'banana', vector: [0.2, 0.1, 0.4]));
index.add(LocalHnswItem(item: 'grape', vector: [0.9, 0.8, 0.7]));

final result = index.search([0.1, 0.2, 0.3], 2);

for (final r in result.items) {
  print('Found ${r.item} with distance ${r.distance}');
}

🔁 Save and Load #

final saved = index.save(encodeItem: (v) => v);
final loaded = LocalHNSW<String>.load(
  json: saved,
  dim: 3,
  decodeItem: (s) => s,
);

📌 API Overview #

LocalHNSW<T> (Generic) #

🧪 Running Tests #

dart test

📄 License #

MIT License