@ruvector/rvlite

Standalone embedded vector database with multi-query-language support (SQL, SPARQL, Cypher), powered by RuVector WASM for zero-dependency portable vector search in any JavaScript runtime.

Quick Command Reference

Task	Code
Create database	`const db = new RVLite()`
Insert vector	`await db.insert(id, vector, metadata)`
Search	`await db.search(queryVector, topK)`
SQL query	`await db.query("SELECT * FROM vectors")`
Cypher query	`await db.query("MATCH (n) RETURN n")`
SPARQL query	`await db.query("SELECT ?s WHERE { ... }")`
Build index	`await db.buildIndex()`
Save to file	`await db.save('./db.rvlite')`

Installation

Hub install (recommended): npx ruvector@latest includes this package. Standalone: npx @ruvector/rvlite@latest See Installation Guide for the full ecosystem.

Core API

RVLite Constructor

typescript

import { RVLite } from '@ruvector/rvlite';

const db = new RVLite({
  dimensions: 384,
  metric: 'cosine',
  persistPath: './my-db.rvlite',
  enableSQL: true,
  enableSPARQL: true,
  enableCypher: true,
});

Constructor Options:

Parameter	Type	Description	Default
`dimensions`	`number`	Vector dimensionality	`384`
`metric`	`string`	Distance metric	`'cosine'`
`persistPath`	`string`	File persistence path	In-memory
`enableSQL`	`boolean`	Enable SQL queries	`true`
`enableSPARQL`	`boolean`	Enable SPARQL queries	`true`
`enableCypher`	`boolean`	Enable Cypher queries	`true`

Vector Operations

typescript

// Insert
await db.insert('doc-1', [0.1, 0.2, 0.3], { title: 'Hello', category: 'greeting' });

// Batch insert
await db.batchInsert(items);

// Search
const results = await db.search(queryVector, 10);

// Search with filter
const results = await db.search(queryVector, 10, { filter: { category: 'greeting' } });

Multi-Language Queries

typescript

// SQL
const result = await db.query("SELECT id, metadata FROM vectors WHERE metadata->>'category' = 'greeting'");

// Cypher
const result = await db.query("MATCH (n:Vector) WHERE n.category = 'greeting' RETURN n");

// SPARQL
const result = await db.query("SELECT ?id WHERE { ?id :category 'greeting' }");

// Auto-detection of query language
const result = await db.query(queryString); // Auto-detects SQL/Cypher/SPARQL

Persistence

typescript

await db.save('./my-db.rvlite');                         // Save to file
const db = await RVLite.load('./my-db.rvlite');          // Load from file
const bytes = db.serialize();                             // Serialize to bytes
const db = RVLite.deserialize(bytes);                     // From bytes

Common Patterns

Embedded Search Engine

typescript

const db = new RVLite({ dimensions: 384, persistPath: './search.rvlite' });
for (const doc of documents) {
  await db.insert(doc.id, doc.embedding, { title: doc.title, text: doc.text });
}
await db.buildIndex();
const results = await db.search(queryEmbedding, 5);

SQL-Based Analytics on Vectors

typescript

const result = await db.query(`
  SELECT metadata->>'category' AS cat, COUNT(*) AS cnt
  FROM vectors
  GROUP BY metadata->>'category'
  ORDER BY cnt DESC
`);

Knowledge Graph with Cypher

typescript

await db.query("CREATE (a:Person {name: 'Alice'})");
await db.query("CREATE (b:Person {name: 'Bob'})");
await db.query("CREATE (a)-[:KNOWS]->(b)");
const friends = await db.query("MATCH (a:Person)-[:KNOWS]->(b) RETURN b.name");

Key Options

Feature	Value
Query languages	SQL, SPARQL, Cypher
Dependencies	None (WASM)
Persistence	File or binary
Runtimes	Node.js, Browser, Deno, Bun
Embedding support	Any (bring your own)

RAN DDD Context

Bounded Context: Data Infrastructure

References

•API reference: See references/commands.md
•Full README
•npm