Haiku-Search


Overview

Haiku-Search is a high-performance fuzzy search library designed for web applications. It is built using Rust and compiled to WebAssembly, providing lightning-fast search capabilities directly in the browser. This combination allows Haiku-Search to execute complex search algorithms efficiently, taking advantage of Rust's performance and safety features.

Powered by Rust and WebAssembly

Haiku-Search leverages the power of Rust and WebAssembly to run directly in web browsers, offering significant performance improvements over traditional JavaScript search libraries. By compiling to WebAssembly, Haiku-Search provides near-native speed, making it ideal for applications requiring quick search responses on large datasets.

Algorithms

Haiku-Search implements two primary algorithms:
- N-gram Indexing: This technique breaks down text into chunks (n-grams) and indexes them for quick lookup, allowing us to quickly narrow down potential matches.
- Bitap Algorithm: For precise matching, Haiku-Search uses the Bitap algorithm, which supports a configurable number of errors. It is effective for short to medium text lengths and allows for approximate string matching.

Installation

WIP
Since Haiku-Search is compiled to WebAssembly and not published on NPM, it can be included in your project by directly linking to the generated WebAssembly and JavaScript loader files. Here is how you can include it in your web project:
``html

`

Usage Example

Here’s a quick example on how to use Haiku-Search in a web application:
`javascript
import init, { SearchConfig, SearchEngine } from './path/to/haiku_pkg/haiku.js';

async function performSearch() {
await init();
const config = new SearchConfig(3, 1); // ngram size and max allowed errors
const engine = new SearchEngine(["search me", "search me not"], config);
const results = engine.search("search");
console.log(results);
}
performSearch();
``

Performance Comparison to Fuse.js

Haiku-Search is designed to be significantly faster than traditional JavaScript libraries like Fuse.js). In benchmarks, Haiku-Search performs searches up to 13x faster than Fuse.js.

!image

You can see this chart live here: Haiku-Search vs Fuse.js Benchmark Results.

Known Limitations

- Unicode Support: Haiku-Search does not currently support unicode characters, which may limit its use in applications requiring internationalization.
- Pattern Length: The Bitap algorithm used in Haiku-Search supports patterns up to 32 characters long due to limitations in handling bit masks.

Roadmap

- Add unicode support
- Improve memory footprint
- Improve code documentation

Contributing

Contributions to Haiku-Search are welcome! If you're looking to contribute, please:
- Check out the current issues on GitHub, especially those tagged as "good first issue".
- Fork the repository and make your changes.
- Write clear, commented code.
- Ensure your changes pass all existing tests and add new tests for added functionality.
- Submit a pull request with a detailed description of your changes.

For major changes, please open an issue first to discuss what you would like to change.

Thank you for your interest in improving Haiku-Search!