Lunr.js — Full-Text Search Engine for the Browser

Introduction

Lunr.js is a small, full-text search library for use in the browser and Node.js. Inspired by Solr, it provides a simple way to build a search index over your documents and query it with features like stemming, boosting, and wildcards, all without external services.

What Lunr.js Does

• Builds a full-text search index from JavaScript objects with configurable fields
• Supports stemming, stop word removal, and language-specific text processing
• Enables field boosting to prioritize matches in specific document fields
• Provides wildcard, fuzzy matching, and term presence/absence operators
• Serializes indexes to JSON for pre-building at compile time and loading at runtime

Architecture Overview

Lunr.js constructs an inverted index during the build phase, tokenizing and stemming each document field into terms that map to document references. The index uses a trie-based structure for efficient prefix and wildcard lookups. At query time, the search engine resolves each query clause against the index, scores matches using TF-IDF with optional field boosts, and returns sorted results.

Self-Hosting & Configuration

• Install via npm or include via a script tag from a CDN
• Define an index inside the lunr callback, specifying fields and adding documents
• Set field boosts with this.field(name, {boost: N}) to weight certain fields higher
• Pre-build the index server-side and serialize with JSON.stringify for faster client-side loading
• Add language support via lunr-languages plugin packs for stemming and stop words

Key Features

• Zero dependencies and works identically in Node.js and browsers
• Pre-built index serialization to JSON for static sites and documentation
• Language support for 14+ languages via community stemmer plugins
• Query syntax with wildcards, fuzzy matches, boosts, and required/prohibited terms
• Pipeline-based text processing that is customizable with user-defined functions

Comparison with Similar Tools

• Fuse.js — fuzzy search without indexing; better for small datasets, no stemming or field boosting
• FlexSearch — faster for very large datasets with context-based scoring; less standard query syntax
• MiniSearch — similar API with auto-suggest and prefix search; actively maintained alternative
• Algolia — hosted search-as-a-service with dashboards and analytics; requires a server and API key

FAQ

Q: Can I update the index after building it? A: No, Lunr indexes are immutable once built. Rebuild the index when your data changes.

Q: How large a dataset can Lunr handle? A: It works well for up to tens of thousands of documents. Larger datasets may cause slow build times or high memory use.

Q: Can I use Lunr with TypeScript? A: Yes, type definitions are available via @types/lunr on npm.

Q: How do I search across multiple fields at once? A: All indexed fields are searched by default. Use field-prefixed queries like title:keyword to restrict to a specific field.

Sources

• https://github.com/olivernn/lunr.js
• https://lunrjs.com/