PeliSearch

Lightweight, embeddable search engine built for modern applications.

Full-text search, filtering, aggregations, and vector search in a single developer-friendly engine.

---

Why PeliSearch?

Modern search engines are powerful, but often come with significant operational complexity.

PeliSearch is designed around a simple idea:

Search should be easy to embed, easy to deploy, and easy to understand.

Whether you're building a SaaS application, an internal tool, a documentation site, or a local desktop application, PeliSearch provides fast search capabilities without requiring a large distributed cluster.

Core Principles

• Embeddable-first architecture
• Single-node simplicity
• High-performance search
• Modern developer experience
• Open-source and Apache 2.0 licensed
• Built for lexical and semantic search

---

Features

Full-Text Search

Fast and relevant search powered by BM25 ranking.

{
  "query": {
    "match": {
      "title": "electric bike"
    }
  }
}

Filtering

Filter results using exact matches, ranges, and dates.

{
  "filter": {
    "range": {
      "price": {
        "lte": 1500
      }
    }
  }
}

Aggregations

Generate analytics and faceted search experiences.

• Terms aggregations
• Count
• Min
• Max
• Average
• Sum

Vector Search (Roadmap)

(Not yet implemented)

Semantic search powered by embeddings and vector similarity.

Hybrid Search (Roadmap)

(Not yet implemented)

Combine lexical and vector search using modern ranking strategies.

---

Quick Example

Create an Index

import { PeliSearchClient } from '@pelisearch/client';

const client = new PeliSearchClient({ host: "http://localhost:7700" });

await client.indexes.create("products");

Index Documents

await client.documents.add("products", {
  id: "1",
  title: "Electric Bike",
  price: 999
});

Search

const results = await client.search.search("products", {
  query: {
    match: {
      title: "bike"
    }
  }
});

---

Architecture

Applications
      │
      ▼
 SDK Layer
      │
      ▼
 Search Engine
      │
 ┌────┼────┐
 ▼    ▼    ▼
Text Filter Ranking
      │
      ▼
 Storage Engine
      │
      ▼
 WAL + Segments

---

Supported SDKs

Available SDKs:

• Rust
• Node.js
• Python
• Go

All SDKs expose a consistent API.

---

Documentation

Getting Started

• Installation
• Quick Start
• First Index
• First Search

Concepts

• Architecture
• Documents
• Indexes
• Mappings
• Storage Engine
• Search Ranking

Guides

• Full-Text Search
• Filtering
• Sorting
• Pagination
• Aggregations
• Vector Search
• Hybrid Search

Architecture

• Storage Engine
• Inverted Index
• Query Engine
• Ranking Engine
• Vector Engine

See the full documentation in the /docs directory.

---

Roadmap

Version 1

• Single-node architecture
• Full-text search
• Filtering
• Sorting
• Aggregations
• Embedded SDKs

Version 2

• Vector search
• Hybrid search
• Advanced ranking

Future

• Replication
• Clustering
• Distributed search

---

Status

PeliSearch is currently under active development.

The initial release focuses on building a fast, embeddable search engine with a strong foundation before expanding into vector and distributed search capabilities.

---

License

Apache License 2.0

Copyright (c) PeliSearch Contributors