Laurus : Lexical Augmented Unified Retrieval Using Semantics

Laurus is a search platform written in Rust — built for Lexical Augmented Unified Retrieval Using Semantics. Built on a core library covering lexical search, vector search, and hybrid search, it provides multiple ready-to-use interfaces:

• Core Library — Modular search engine embeddable into any application
• CLI & REPL — Command-line tool for interactive search experiences
• gRPC Server & HTTP Gateway — Seamless integration with microservices and existing systems
• MCP Server — Direct integration with AI assistants such as Claude
• Python Bindings — Native Python package for use in data science and AI workflows

Whether embedded as a library, deployed as a standalone server, called from Python, or woven into AI workflows, Laurus is a composable search foundation.

Documentation

Comprehensive documentation is available online:

• English: https://mosuka.github.io/laurus/
• Japanese (日本語): https://mosuka.github.io/laurus/ja/

Contents

• Getting Started
  • Installation
  • Quick Start
  • Examples
• Core Concepts
  • Schema & Fields
  • Text Analysis
  • Embeddings
  • Storage
  • Indexing (Lexical / Vector)
  • Search (Lexical / Vector / Hybrid)
  • Query DSL
• Crate Guides
  • laurus (Library) — Engine, Scoring, Faceting, Highlighting, Spelling Correction, Persistence & WAL
  • laurus-cli — Command-line interface, REPL, Schema Format
  • laurus-server — gRPC server, HTTP Gateway, Configuration
  • laurus-mcp — MCP server for AI assistants (Claude, etc.)
  • laurus-python — Python bindings (PyPI package)
• Development
  • Build & Test
  • Feature Flags
  • Project Structure
• API Reference (docs.rs)

Features

• Pure Rust Implementation: Memory-safe and fast performance with zero-cost abstractions.
• Hybrid Search: Seamlessly combine BM25 lexical search with HNSW vector search using configurable fusion strategies.
• Multimodal Capabilities: Native support for text-to-image and image-to-image search via CLIP embeddings.
• Rich Query DSL: Term, phrase, boolean, fuzzy, wildcard, range, geographic, and span queries.
• Flexible Analysis: Configurable pipelines for tokenization, normalization, and stemming (including CJK support via Lindera).
• Pluggable Storage: Interfaces for in-memory, file-system, and memory-mapped storage backends.
• Scoring & Ranking: BM25 scoring with customizable fusion strategies for hybrid results.
• Faceting & Highlighting: Built-in support for faceted navigation and search result highlighting.
• Spelling Correction: Suggest corrections for misspelled query terms.

Workspace Structure

Laurus is organized as a Cargo workspace with 5 crates:

Crate	Description
laurus	Core search library — schema, analysis, indexing, search, and storage
laurus-cli	Command-line interface with REPL for interactive search
laurus-server	gRPC server with HTTP gateway for deploying Laurus as a service
laurus-mcp	MCP server for AI assistants (Claude, etc.) via stdio transport
laurus-python	Python bindings (PyPI package) built with PyO3 and Maturin

Feature Flags

The laurus crate provides optional feature flags for embedding support:

Feature	Description
embeddings-candle	Local BERT embeddings via Candle
embeddings-openai	Cloud-based embeddings via the OpenAI API
embeddings-multimodal	CLIP-based multimodal (text + image) embeddings
embeddings-all	Enable all embedding backends

Quick Start

use laurus::lexical::{TermQuery, TextOption};
use laurus::storage::memory::MemoryStorageConfig;
use laurus::storage::{StorageConfig, StorageFactory};
use laurus::{Document, Engine, LexicalSearchRequest, Schema, SearchRequestBuilder};

#[tokio::main]
async fn main() -> laurus::Result<()> {
    // 1. Create storage
    let storage = StorageFactory::create(StorageConfig::Memory(MemoryStorageConfig::default()))?;

    // 2. Define schema
    let schema = Schema::builder()
        .add_text_field("title", TextOption::default())
        .add_text_field("body", TextOption::default())
        .build();

    // 3. Create engine
    let engine = Engine::new(storage, schema).await?;

    // 4. Index documents
    engine
        .add_document(
            "doc1",
            Document::builder()
                .add_text("title", "Introduction to Rust")
                .add_text(
                    "body",
                    "Rust is a systems programming language focused on safety and performance.",
                )
                .build(),
        )
        .await?;
    engine
        .add_document(
            "doc2",
            Document::builder()
                .add_text("title", "Python for Data Science")
                .add_text(
                    "body",
                    "Python is a versatile language widely used in data science and machine learning.",
                )
                .build(),
        )
        .await?;
    engine.commit().await?;

    // 5. Search
    let results = engine
        .search(
            SearchRequestBuilder::new()
                .lexical_search_request(LexicalSearchRequest::new(Box::new(TermQuery::new(
                    "body", "rust",
                ))))
                .limit(5)
                .build(),
        )
        .await?;

    for hit in &results {
        println!("score={:.4}", hit.score);
    }

    Ok(())
}

Examples

You can find usage examples in the laurus/examples/ directory:

Example	Description	Feature Flag
quickstart	Basic full-text search	—
lexical_search	All query types (Term, Phrase, Boolean, Fuzzy, Wildcard, Range, Geo, Span)	—
vector_search	Semantic similarity search with embeddings	—
hybrid_search	Combining lexical and vector search with fusion	—
synonym_graph_filter	Synonym expansion in analysis pipeline	—
search_with_candle	Local BERT embeddings via Candle	embeddings-candle
search_with_openai	Cloud-based embeddings via OpenAI	embeddings-openai
multimodal_search	Text-to-image and image-to-image search	embeddings-multimodal

Contributing

We welcome contributions!

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add some amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

License

This project is licensed under the MIT License - see the LICENSE file for details.