Scala implementation of Aho-Corasick algorithm

A byte-oriented Aho-Corasick implementation in pure Scala. Keywords can carry a user-defined property tag.

Features

• UTF-8 is used explicitly for string APIs, so behavior is consistent across platforms
• Failure links are built automatically on the first search (or call build() eagerly)
• Array-based goto table and vocabulary indexing for better performance and memory use on large dictionaries

Requirements

• Java 11+
• Maven 3.6+

Build and test

mvn test

Usage

import io.yizhiru.ac.Automaton

val ac = new Automaton
ac.addWords("pronoun", "he", "she", "his", "hers")

// build() is optional; search triggers construction automatically
val results: Set[(String, String)] = ac.search("ushers")
// Set(("he", "pronoun"), ("she", "pronoun"), ("hers", "pronoun"))

Search raw bytes when you already have UTF-8 (or other byte-level) data:

ac.addWordBytes("tag", "he".getBytes(java.nio.charset.StandardCharsets.UTF_8))
ac.searchBytes("ushers".getBytes(java.nio.charset.StandardCharsets.UTF_8))

API notes

Method	Description
addWord(property, word)	Register one UTF-8 keyword
addWords(property, words*)	Register multiple keywords with the same property
addWordBytes(property, bytes)	Register a keyword from raw bytes
build()	Eagerly construct failure links and complete the goto table
search(text)	Match UTF-8 text, returns Set[(keyword, property)]
searchBytes(data)	Match raw bytes

setFailTransitions() is deprecated and kept only for backward compatibility; it delegates to build().

Duplicate matches of the same (keyword, property) pair in one text are deduplicated because results are returned as a Set.

Encoding

All string-based methods encode and decode with StandardCharsets.UTF_8. If your input uses another charset, convert to bytes yourself and use addWordBytes / searchBytes.