Skip to main content

Levenshtein edit distance library

Project description

Levenshtein edit distance library for Python, Apache-licensed. Written by Lars Buitinck, Netherlands eScience Center, with contributions from Isaac Sijaranamual, University of Amsterdam.

Performs distance computations on either byte strings or Unicode codepoints.

Installation

Make sure you have Cython and a C++ compiler installed:

pip install cython

Installing a C++ compiler is so platform-dependent that I won’t show instructions. Consult your package manager.

Then:

python setup.py install

To run the tests, but not to actually use leven, you need six and Nose.

Usage

>>> from leven import levenshtein
>>> levenshtein("hello, world!", "goodbye, cruel world!")
13

About the implementation

The core algorithms have been implemented in C++. I used this instead of C to get templates, easier memory management and a better standard library, so the C++ code probably looks C-ish.

Todo

  • Implement Ukkonen’s algorithm for bounded Levenshtein distance

  • Implement Levenshtein automata for fast neighbor search in string spaces

  • Implement weighted Levenshtein distance

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page