29 releases (stable)
2.2.2 | Dec 28, 2022 |
---|---|
2.2.0 | Dec 15, 2022 |
2.1.3 | Aug 26, 2022 |
2.1.2 | Jul 24, 2022 |
0.1.15 | May 20, 2022 |
#265 in Algorithms
11,655 downloads per month
Used in 10 crates
(4 directly)
50KB
721 lines
Stringmetrics
This is a Rust library for approximate string matching that implements simple algorithms such has Hamming distance, Levenshtein distance, Jaccard similarity, and more.
Here are some useful quick links:
- Crate info: https://crates.io/crates/stringmetrics
- Crate docs: https://docs.rs/stringmetrics/
- Python library page: https://pypi.org/project/stringmetrics/
- Crate source: https://github.com/pluots/stringmetrics
Algorithms
The main purpose of this library is to provide a variety of string metric functions. Included algorithms are:
- Levenshtein Distance
- Limited & Weighted Levenshtein Distance
- Jaccard Similarity
- Hamming Distance
See the documentation for full information. Some examples are below:
// Basic levenshtein distance
use stringmetrics::levenshtein;
assert_eq!(levenshtein("kitten", "sitting"), 3);
// Levenshtein distance with a limit to save computation time
use stringmetrics::levenshtein_limit;
assert_eq!(levenshtein_limit("a very long string", "short!", 4), 4);
// Set custom weights
use stringmetrics::{levenshtein_weight, LevWeights};
// This struct holds insertion, deletion, and substitution costs
let weights = LevWeights::new(4, 3, 2);
assert_eq!(levenshtein_weight("kitten", "sitting", 100, &weights), 8);
// Basic hamming distance
use stringmetrics::hamming;
let a = "abcdefg";
let b = "aaadefa";
assert_eq!(hamming(a, b), Ok(3));
Future Algorithms & Direction
Eventually, this library aims to add support for more algorithms. Intended work includes:
- Update levenshtein distance to have a more performant algorithm for short (<64 characters) and long (>100 characters) strings
- Add the Damerau–Levenshtein distance
- Add the Jaro–Winkler distance
- Add the Tversky index
- Add Cosine similarity
- Add some useful tokenizers with examples
License
See the LICENSE file for license information. The provided license does allow for proprietary use and adaptation; that being said, I kindly suggest that if you come up with an improvement, you submit a pull request and help us all out :)