Fast, RapidFuzz-compatible fuzzy string matching for Kotlin Multiplatform.

• Verified parity — bit-exact with RapidFuzz, checked against 3,260 golden vectors (scores and scoreCutoff behavior) on every commit. Tune thresholds in a Python notebook, ship the numbers to Kotlin unchanged.
• Bit-parallel core — Hyyrö's bit-parallel LCS packs the DP column into machine words; extraction scans reuse the query's match masks across the whole collection and short-circuit on scoreCutoff.
• Correct on full Unicode — all algorithms operate on code points, never UTF-16 chars. A single emoji is one symbol, astral-plane characters sort correctly, and defaultProcess uses the exact Unicode tables of RapidFuzz's C++ backend.
• Every KMP target — JVM (serves Android), JS, Wasm, iOS (including iosX64), macOS, watchOS, tvOS, Linux, Windows, Android Native.
• Zero runtime dependencies, MIT licensed.

Name note: several unrelated projects are also called "kmatch"/"KMatcher" (including a Kotlin pattern-matching DSL). This library is fuzzy string matching, has no DSL, and lives at the Maven coordinate io.github.likhithsj:kmatch — when in doubt, that coordinate is the identity.

Live playground → — compare strings across every scorer and search a collection with match highlighting, running the real library as Kotlin/JS in your browser. API docs → (Source in demo/.)

implementation("io.github.likhithsj:kmatch:0.3.1")

One dependency line, every target: JVM/Android, iOS (device + both simulators), macOS, watchOS, tvOS, JS, Wasm, Linux, Windows, Android Native — Gradle selects the right artifact automatically. The JVM artifact targets Java 8 bytecode, so it runs on any JVM from 8 up. Consuming from Swift? See the SKIE sample.

All scorers return a Double on RapidFuzz's 0–100 scale and match its output exactly:

import io.github.likhithsj.kmatch.Fuzz
import io.github.likhithsj.kmatch.defaultProcess

Fuzz.ratio("this is a test", "this is a test!")        // 96.55172413793103
Fuzz.partialRatio("this is a test", "this is a test!") // 100.0
Fuzz.tokenSortRatio("fuzzy wuzzy was a bear", "wuzzy fuzzy was a bear") // 100.0
Fuzz.tokenSetRatio("fuzzy was a bear but not a dog", "fuzzy was a bear") // 100.0
Fuzz.weightedRatio("this is a test", "this is a new test!!")  // WRatio
Fuzz.quickRatio("this is a test", "this is a new test!!")     // QRatio

// RapidFuzz's default preprocessing: lowercase, non-alphanumeric -> space, trim.
Fuzz.ratio("Test String!", "test string", processor = ::defaultProcess) // 100.0

// scoreCutoff: results below the cutoff return 0.0.
Fuzz.ratio("hello world", "hello world!", scoreCutoff = 99.0) // 0.0

Full surface: ratio, partialRatio, tokenSortRatio, tokenSetRatio, tokenRatio, partialTokenSortRatio, partialTokenSetRatio, partialTokenRatio, weightedRatio (WRatio), quickRatio (QRatio), defaultProcess.

Note that ratio is indel-based (insertions and deletions only), not normalized Levenshtein: ratio("myself", "me") is 50 under indel, 33.3 under Levenshtein. This matches RapidFuzz and fuzzywuzzy semantics and is the single most common porting error.

import io.github.likhithsj.kmatch.*

val choices = listOf("Atlanta Falcons", "New York Jets", "New York Giants", "Dallas Cowboys")

extractOne("new york jets", choices, processor = ::defaultProcess)
// ExtractedResult(choice="New York Jets", score=100.0, index=1)

extractTop("new york", choices, limit = 2, processor = ::defaultProcess)
extractSorted("new york", choices, scoreCutoff = 50.0)
extractAll("new york", choices)

// Custom scorers are a fun interface:
val exact = Scorer { a, b, _ -> if (a == b) 100.0 else 0.0 }
extractOne("query", choices, scorer = exact)

Extraction is generic — score your records directly with a keySelector and get the records back:

data class City(val name: String, val population: Int)

extractOne("new york", cities, keySelector = { it.name }, processor = ::defaultProcess)
// ExtractedItem(item=City(name=New York, population=8300000), score=100.0, index=0)

// Collapse near-duplicates (fuzzywuzzy process.dedupe semantics: each
// duplicate group keeps its longest entry).
dedupe(listOf("Frodo Baggins", "Frodo Baggin", "F. Baggins", "Gandalf"))
// [Frodo Baggins, Gandalf]

// Highlight why a choice matched: char-index ranges into the choice string.
matchingRanges("new york", "the new york times")   // ranges covering "new york"
matchingBlocks("kitten", "sitting")                // aligned blocks in both strings

matchingBlocks/matchingRanges recover one optimal indel alignment with Hirschberg's algorithm — linear memory, so long inputs are safe. Indices are UTF-16 char units, directly usable with substring or AnnotatedString (astral-plane characters handled correctly).

tools/generate_vectors.py pins RapidFuzz 3.14.5 and emits GoldenVectors.kt — 3,260 (inputs, scorer, expected score) cases covering plain ASCII, accented Latin, astral-plane characters, empty strings, strings past 64 code points, token edge cases, and non-Latin scripts through defaultProcess. CI asserts exact float64 equality on every tested target and regenerates the vectors to catch drift.

defaultProcess and tokenization use Unicode tables probed directly from RapidFuzz's C++ backend (tools/generate_unicode_tables.py) rather than any host platform's Unicode APIs — the backends genuinely differ from both Python's re module and Java's Character (underscore handling, NBSP tokenization, simple vs. full case mapping), and the probe-derived tables make kmatch match what RapidFuzz users actually observe.

The Hyyrö bit-parallel LCS replaces the 0.1.0 code-point DP behind the same frozen API — every output is proven unchanged by the golden vectors plus randomized equivalence tests against the DP (kept in test sources as the reference).

Where the speed shows up, measured by the in-repo benchmark (KMATCH_BENCH=1 ./gradlew jvmTest --tests '*BenchmarkTest', JVM 21, median of 7 rounds):

Collection scans win big because the query's match masks are built once and reused for every choice, and scoreCutoff aborts comparisons whose length difference already rules them out. One-shot calls on short string pairs are microsecond-scale either way; if you score one query against many strings, use the extract* functions rather than looping over Fuzz.ratio yourself. Numbers are indicative (container hardware), and the harness is committed so you can reproduce them on yours.

Same harness (thirdPartyComparison), same data, each library's own API; 2,000 pairs / 20,000-choice scan, JVM 21:

Throughput comparison only — output scales and rounding differ between libraries (kmatch is the one matching RapidFuzz bit-exactly). Both libraries are JVM-test-only dependencies of the benchmark; the shipped artifact stays zero-dependency.

• ✅ 0.1.0–0.3.0 (released) — parity core, bit-parallel performance core, generic extraction, dedupe, matchingRanges, published benchmarks, playground, API docs.
• 0.3.1 (current) — JVM artifact targets Java 8 bytecode, so it loads on any JVM from 8 up.
• 1.0.0 — API freeze after early-adopter feedback. Candidates being weighed: a public prepared-query (CachedScorer) API, ready-made accent-folding processors (searchFold), lazy Sequence extraction.

MIT