mobile_rag_engine 0.4.3

A high-performance, on-device RAG (Retrieval-Augmented Generation) engine for Flutter. Run semantic search completely offline on iOS and Android with HNSW vector indexing.

example/example.md

Mobile RAG Engine - Example

A complete on-device RAG (Retrieval-Augmented Generation) implementation.

Quick Start

import 'package:flutter/services.dart';
import 'package:path_provider/path_provider.dart';
import 'package:mobile_rag_engine/mobile_rag_engine.dart';

void main() async {
  WidgetsFlutterBinding.ensureInitialized();
  
  // 1. Initialize Rust library
  await RustLib.init();
  
  // 2. Load tokenizer
  final dir = await getApplicationDocumentsDirectory();
  await initTokenizer(tokenizerPath: '${dir.path}/tokenizer.json');
  
  // 3. Load ONNX embedding model
  final modelBytes = await rootBundle.load('assets/model.onnx');
  await EmbeddingService.init(modelBytes.buffer.asUint8List());
  
  // 4. Initialize RAG service
  final ragService = SourceRagService(dbPath: '${dir.path}/rag.db');
  await ragService.init();
  
  runApp(MyApp(ragService: ragService));
}

Adding Documents

// Add a document with automatic chunking and embedding
final result = await ragService.addSourceWithChunking(
  'Flutter is Google\'s UI toolkit for building beautiful apps...',
  onProgress: (done, total) => print('Embedding: $done/$total'),
);

print('Created ${result.chunkCount} chunks');

// Rebuild HNSW index after adding documents
await ragService.rebuildIndex();

Advanced: PDF & DOCX Support

Can automatically extract text from PDF and DOCX files.

import 'dart:io';

// Read file bytes
final file = File('path/to/document.pdf');
final bytes = await file.readAsBytes();

// explicit extraction
final text = await extractTextFromPdf(bytes);

// OR auto-detect format
final text = await extractTextFromDocument(bytes);

// Then add to RAG
await ragService.addSourceWithChunking(text, metadata: 'source: document.pdf');

Managing Documents

// Remove a document by ID
await ragService.removeSource(sourceId);

// Get duplicate status when adding
final result = await ragService.addSourceWithChunking(content);
if (result.isDuplicate) {
  print('Document already exists (ID: ${result.sourceId})');
}

Semantic Search

// Search for relevant chunks
final searchResult = await ragService.search(
  'How to build mobile apps?',
  topK: 5,
  tokenBudget: 2000,
);

// Get assembled context for LLM
print('Found ${searchResult.chunks.length} chunks');
print('Context tokens: ${searchResult.context.estimatedTokens}');

// Format prompt for LLM
final prompt = ragService.formatPrompt(
  'How to build mobile apps?',
  searchResult,
);

Batch Embedding

// Embed multiple texts efficiently
final embeddings = await EmbeddingService.embedBatch(
  ['Text 1', 'Text 2', 'Text 3'],
  onProgress: (done, total) => print('Progress: $done/$total'),
);

Performance

Operation	Time	Engine
Tokenization	0.04ms	Rust
HNSW Search	0.3ms	Rust
Embedding	25-100ms	ONNX

See the full example app in the GitHub repository.