Migrating from string-similarity to rapid-fuzzy

string-similarity is deprecated on npm and no longer maintained. rapid-fuzzy’s sorensenDice computes the same Dice coefficient, alongside other distance metrics and fuzzy search, all powered by a Rust core.

Installation

# Remove string-similarity
npm uninstall string-similarity

# Install rapid-fuzzy
npm install rapid-fuzzy

API Mapping

string-similarity	rapid-fuzzy	Notes
compareTwoStrings(a, b)	sorensenDice(a, b)	Same Dice coefficient (whitespace removed, case-sensitive)
findBestMatch(s, targets).ratings	sorensenDiceMany(s, targets)	All scores, in the order of targets
findBestMatch(s, targets).bestMatch	sorensenDiceMany + pick the maximum	See Finding the best match

sorensenDice returned the same scores as compareTwoStrings in our differential tests on text without emoji or other characters outside the Basic Multilingual Plane. For those, the results can differ, because rapid-fuzzy counts Unicode code points and string-similarity counts UTF-16 code units ('a😀b' vs 'a😀c': 0.5 vs 0.67).

Code Examples

Comparing two strings

// Before (string-similarity)
import stringSimilarity from 'string-similarity';
const score = stringSimilarity.compareTwoStrings('healed', 'sealed');
// 0.8

// After (rapid-fuzzy)
import { sorensenDice } from 'rapid-fuzzy';
const score = sorensenDice('healed', 'sealed');
// 0.8

Finding the best match

// Before (string-similarity)
import stringSimilarity from 'string-similarity';
const result = stringSimilarity.findBestMatch('healed', ['sealed', 'healthy', 'help']);
console.log(result.bestMatch.target); // 'sealed'
console.log(result.bestMatch.rating); // 0.8

// After (rapid-fuzzy)
import { sorensenDiceMany } from 'rapid-fuzzy';
const targets = ['sealed', 'healthy', 'help'];
const ratings = sorensenDiceMany('healed', targets); // [0.8, 0.5455, 0.25]
const bestIndex = ratings.indexOf(Math.max(...ratings));
console.log(targets[bestIndex]); // 'sealed'
console.log(ratings[bestIndex]); // 0.8

Do not replace findBestMatch with closest(). rapid-fuzzy’s closest() is a fuzzy search: it returns an item that contains the query’s characters in order, and null if there is none. closest('healed', ['sealed', 'healthy', 'help']) returns null.

Getting all similarity scores

// Before (string-similarity)
import stringSimilarity from 'string-similarity';
const result = stringSimilarity.findBestMatch('healed', ['sealed', 'healthy', 'help']);
const ratings = result.ratings.map((r) => r.rating);

// After (rapid-fuzzy)
import { sorensenDiceMany } from 'rapid-fuzzy';
const scores = sorensenDiceMany('healed', ['sealed', 'healthy', 'help']);
// [0.8, 0.5455, 0.25] — scores in the same order as the input array

Batch comparisons

// Before (string-similarity) — no batch API, loop required
import stringSimilarity from 'string-similarity';
const pairs = [['healed', 'sealed'], ['hello', 'world']];
const scores = pairs.map(([a, b]) => stringSimilarity.compareTwoStrings(a, b));

// After (rapid-fuzzy) — batch API
import { sorensenDiceBatch } from 'rapid-fuzzy';
const scores = sorensenDiceBatch([['healed', 'sealed'], ['hello', 'world']]);
// [0.8, 0]

Additional Algorithms

rapid-fuzzy provides algorithms that string-similarity does not:

Algorithm	Function	Best for
Levenshtein distance	levenshtein(a, b)	Typo detection, spell checking
Normalized Levenshtein	normalizedLevenshtein(a, b)	Length-independent comparison
Jaro-Winkler	jaroWinkler(a, b)	Name / address matching
Damerau-Levenshtein	damerauLevenshtein(a, b)	Transposition-aware edit distance
Fuzzy search	search(query, items)	Interactive search / autocomplete

Performance

From the benchmarks (Node.js 22, Linux x64, Intel Xeon @ 2.10GHz; 6 string pairs per operation):

Operation	rapid-fuzzy	string-similarity
Sorensen-Dice (single pairs)	178,705 ops/s	56,320 ops/s

sorensenDice was about 3x faster than compareTwoStrings. sorensenDiceMany prepares the reference string once for all candidates.

Key Differences

• Return values: Both libraries return 0.0–1.0 similarity scores for the Dice coefficient.
• findBestMatch: rapid-fuzzy has no single equivalent; use sorensenDiceMany() and pick the maximum, as shown above.
• TypeScript: rapid-fuzzy ships with built-in TypeScript declarations. No @types/ package needed.
• ESM/CJS: rapid-fuzzy supports both ESM (import) and CommonJS (require).