Matchete

Matchete is a versatile Rust library for similarity matching between query and candidate items. Designed for flexibility, it enables custom similarity metrics, weighted metric combinations, and configurable matching behavior via a builder pattern. Whether you're building search engines, recommendation systems, or fuzzy matching tools, Matchete provides a robust, generic framework that works with any data types implementing Clone and Debug.

The name Matchete draws inspiration from "machete," evoking precision and strength in cutting through complex matching tasks, while playfully nodding to its core purpose: finding the best match.

Features

• Generic Matching: Supports any query and candidate types with Clone and Debug traits.
• Custom Metrics: Define tailored similarity metrics via the SimilarityMetric trait.
• Weighted Scoring: Combine multiple metrics with adjustable weights.
• Composite Matching: Use MultiMatcher to aggregate results from multiple matchers.
• Flexible Configuration: Set thresholds, conditional rules, and options with MatcherBuilder.
• Detailed Insights: Access per-metric scores and match types for in-depth analysis.
• Extensible: Easily add new metrics or strategies to suit your needs.

Installation

Add Matchete to your project by including it in your Cargo.toml:

[dependencies]
matchete = "0.0.1"  # Replace with the actual version

Ensure you have Rust installed via rustup.

Usage

Matchete offers a simple yet powerful API for matching tasks. Below is a basic example of string matching with a custom metric. For more examples, see the Examples section.

use matchete::{Matcher, SimilarityMetric};

// Define a simple string similarity metric
struct SimpleMetric;

impl SimilarityMetric<String, String> for SimpleMetric {
    fn calculate(&self, query: &String, candidate: &String) -> f64 {
        if query == candidate { 1.0 } else { 0.5 }
    }

    fn id(&self) -> &str { "simple" }
}

fn main() {
    // Create a matcher
    let matcher = Matcher::<String, String>::new()
        .with_metric(SimpleMetric, 1.0)
        .with_threshold(0.6);

    // Match a query against candidates
    let query = String::from("hello");
    let candidates = vec![String::from("hello"), String::from("world")];
    if let Some(result) = matcher.find_best_match(&query, &candidates) {
        println!("Best match: {} (score: {:.2})", result.candidate, result.score);
    }
}

Examples

The examples folder contains practical demonstrations of Matchete's capabilities:

• basic_string_matching.rs: Simple string matching with a single metric.
• multiple_metrics.rs: Combining metrics using MatcherBuilder with conditional thresholds.
• numeric_matching.rs: Matching f64 numbers with a custom metric.
• detailed_analysis.rs: Analyzing matches with per-metric score details.
• multi_matcher.rs: Combining multiple matchers with MultiMatcher.

Run an example using:

cargo run --example basic_string_matching

Each example includes detailed comments and instructions. The examples/common.rs file provides shared metrics used across these demos.

Configuration

Customize Matchete's behavior with:

• Thresholds: Set minimum scores for matches (default: 0.4).
• Weights: Assign importance to each metric.
• Options: Configure settings like case sensitivity via key-value pairs.
• Conditional Thresholds: Apply dynamic thresholds based on query properties.
• Metric Groups: Organize metrics for advanced strategies like fallback chains.

See multiple_metrics.rs for an example of complex configurations.

Contributing

Contributions are welcome! To contribute:

1. Fork the repository.
2. Create a branch for your feature or fix (git checkout -b feature/my-feature).
3. Commit your changes (git commit -m "Add my feature").
4. Push to your branch (git push origin feature/my-feature).
5. Open a pull request with a clear description.

Please include tests and update documentation. Adhere to the Rust Code of Conduct.

License

Matchete is licensed under the MIT License. See the LICENSE file for details.

Contact

For questions, suggestions, or issues, visit the GitHub repository.