Set up documentation site using MkDocs and mkdocstrings

Author: maskedsyntax

docs/api.md

@@ -0,0 +1,21 @@
+# API Reference
+
+## Basic Matching
+
+::: fuzzybunny.levenshtein
+::: fuzzybunny.partial_ratio
+::: fuzzybunny.jaccard
+::: fuzzybunny.token_sort
+::: fuzzybunny.token_set
+::: fuzzybunny.qratio
+::: fuzzybunny.wratio
+
+## Ranking
+
+::: fuzzybunny.rank
+::: fuzzybunny.batch_match
+
+## Utilities
+
+::: fuzzybunny.benchmark.benchmark
+::: fuzzybunny.benchmark.benchmark_batch

docs/index.md

@@ -0,0 +1,36 @@
+# FuzzyBunny
+
+A high-performance, lightweight Python library for fuzzy string matching and ranking, implemented in C++ with Pybind11.
+
+## Features
+
+- **Blazing Fast**: Optimized C++ core (Myers' Bit-Parallel algorithm) for superior performance.
+- **Multiple Scorers**: Support for Levenshtein, Jaccard, Token Sort, Token Set, QRatio, and WRatio.
+- **Partial Matching**: Find the best substring matches.
+- **Hybrid Scoring**: Combine multiple scorers with custom weights.
+- **Python Callbacks**: Use your own Python functions as scorers.
+- **Pandas & NumPy Integration**: Native support for Series and Arrays.
+- **Parallelized**: Parallel matching for large datasets using OpenMP.
+
+## Quick Start
+
+```python
+import fuzzybunny
+
+# Basic matching
+score = fuzzybunny.levenshtein("kitten", "sitting")
+print(f"Similarity: {score:.2f}")
+
+# Ranking candidates
+candidates = ["apple", "apricot", "banana", "cherry"]
+results = fuzzybunny.rank("app", candidates, top_n=2)
+# [('apple', 0.6), ('apricot', 0.42)]
+```
+
+## Installation
+
+```bash
+pip install fuzzybunny
+```
+
+*Note: On macOS, it is recommended to have `libomp` installed via Homebrew for full parallel processing support.*

docs/performance.md

@@ -0,0 +1,28 @@
+# Performance Benchmarks
+
+FuzzyBunny is designed for speed, utilizing an optimized C++ core with bit-parallel algorithms and multi-threading support.
+
+## Core Matching Performance
+
+The core Levenshtein implementation uses **Myers' Bit-Parallel algorithm** for strings up to 64 characters. This allows for $O(N)$ matching with very low constant overhead.
+
+For longer strings, it falls back to an optimized $O(NM)$ dynamic programming approach.
+
+## Multi-threading with OpenMP
+
+`batch_match` is parallelized using **OpenMP**. This allows you to process large numbers of queries across all available CPU cores without being bottlenecked by the Python Global Interpreter Lock (GIL).
+
+## In-place Normalization
+
+By pre-normalizing candidates and queries in `batch_match`, we avoid redundant computations, making it significantly faster for bulk operations compared to repetitive calls to individual matchers.
+
+## Benchmarking on Your Machine
+
+You can run the built-in benchmarking tool to see how it performs on your specific hardware and data:
+
+```python
+import fuzzybunny
+
+perf = fuzzybunny.benchmark("query", ["candidate"] * 10000)
+print(perf)
+```

mkdocs.yml

@@ -0,0 +1,20 @@
+site_name: FuzzyBunny
+site_url: https://cachevector.github.io/fuzzybunny/
+repo_url: https://github.com/cachevector/fuzzybunny
+theme:
+  name: material
+  palette:
+    primary: deep purple
+    accent: pink
+
+plugins:
+  - search
+  - mkdocstrings:
+      handlers:
+        python:
+          paths: [src]
+
+nav:
+  - Home: index.md
+  - API Reference: api.md
+  - Performance: performance.md