Crate kiwi_rs 

Rust bindings for Kiwi C API.

This crate provides a high-level API that is convenient for day-to-day use, while still exposing lower-level handles for advanced scenarios.

Quick Start

use kiwi_rs::Kiwi;

fn main() -> Result<(), Box<dyn std::error::Error>> {
    let kiwi = Kiwi::init()?;
    let tokens = kiwi.tokenize("아버지가방에들어가신다.")?;
    for token in tokens {
        println!("{}/{}", token.form, token.tag);
    }
    Ok(())
}

Initialization Paths

kiwi-rs supports two common initialization modes:

1. Automatic bootstrap via Kiwi::init

• Uses local paths first.
• If unavailable, downloads a matching Kiwi library/model pair into cache.

2. Explicit setup via Kiwi::from_config

• For controlled deployments with fixed library/model paths.

use kiwi_rs::{Kiwi, KiwiConfig};

fn main() -> Result<(), Box<dyn std::error::Error>> {
    let config = KiwiConfig::default()
        .with_library_path("/path/to/libkiwi.dylib")
        .with_model_path("/path/to/models/cong/base");
    let kiwi = Kiwi::from_config(config)?;
    let _tokens = kiwi.tokenize("형태소 분석 예시")?;
    Ok(())
}

Offset And Unit Rules

• For UTF-8 APIs, offsets are character indices (based on str.chars()), not byte indices.
• UTF-16 APIs accept &[u16], but returned text in this crate is converted back to Rust UTF-8 String.

Environment Variables

• KIWI_LIBRARY_PATH: explicit dynamic library path.
• KIWI_MODEL_PATH: explicit model directory path.
• KIWI_RS_VERSION: version used by Kiwi::init bootstrap (latest by default).
• KIWI_RS_CACHE_DIR: cache root used by Kiwi::init bootstrap.

Structs

AnalysisCandidate

One analysis candidate, including probability and token list.

AnalyzeOptions

Options for analyze* and tokenize* APIs.

BuilderConfig

Builder-time configuration for constructing a crate::Kiwi instance.

ExtractedWord

Candidate extracted word from extract_words* builder APIs.

GlobalConfig

Global runtime parameters for Kiwi inference behavior.

Kiwi

High-level Kiwi analyzer instance.

KiwiBuilder

Builder used to configure dictionaries/rules and then construct Kiwi.

KiwiConfig

Top-level configuration used by crate::Kiwi::from_config.

KiwiLibrary

Handle to a loaded Kiwi dynamic library plus resolved function table.

KiwiTypo

Typo model/preset handle used when building Kiwi.

MorphemeInfo

Morpheme metadata from dictionary lookup APIs.

MorphemeSense

Morpheme information with resolved string fields.

MorphemeSet

Morpheme id set used as a blocklist in analysis/tokenization APIs.

PreAnalyzedToken

Pre-analyzed token element passed to crate::KiwiBuilder::add_pre_analyzed_word.

PreparedJoinMorphs

Reusable join input for high-throughput join calls.

PreparedJoiner

Reusable joiner handle bound to a specific morph sequence.

Pretokenized

Container for user-supplied token spans used during analysis overrides.

Sentence

Sentence split result used by split_into_sents*_with_options.

SentenceBoundary

Begin/end boundary for a sentence in character offsets.

SimilarityPair

(id, score) pair returned by similarity and prediction APIs.

SwTokenizer

Subword tokenizer model handle opened from Kiwi-compatible tokenizer files.

Token

A single morpheme token produced by Kiwi analysis.

TokenInfo

Low-level token metadata returned by Kiwi C API.

UserWord

A user dictionary entry consumed by crate::KiwiBuilder::add_user_words.

Enums

KiwiError

Error type returned by kiwi-rs public APIs.

Constants

KIWI_BUILD_DEFAULT

Default build option mask.

KIWI_BUILD_DEFAULT_WITH_CONG

Default build options with CoNg model.

KIWI_BUILD_INTEGRATE_ALLOMORPH

Build option: integrate allomorph variants.

KIWI_BUILD_LOAD_DEFAULT_DICT

Build option: load bundled default dictionary.

KIWI_BUILD_LOAD_MULTI_DICT

Build option: load multi-word dictionary.

KIWI_BUILD_LOAD_TYPO_DICT

Build option: load typo dictionary.

KIWI_BUILD_MODEL_TYPE_CONG

Build option: CoNg model type.

KIWI_BUILD_MODEL_TYPE_CONG_GLOBAL

Build option: global CoNg model type.

KIWI_BUILD_MODEL_TYPE_DEFAULT

Build option: default model type.

KIWI_BUILD_MODEL_TYPE_KNLM

Build option: KNLM model type.

KIWI_BUILD_MODEL_TYPE_LARGEST

Build option: largest model type.

KIWI_BUILD_MODEL_TYPE_SBG

Build option: SBG model type.

KIWI_DIALECT_ALL

Dialect mask containing all supported dialect flags.

KIWI_DIALECT_ARCHAIC

Dialect flag: archaic expressions.

KIWI_DIALECT_CHUNGCHEONG

Dialect flag: Chungcheong.

KIWI_DIALECT_GANGWON

Dialect flag: Gangwon.

KIWI_DIALECT_GYEONGGI

Dialect flag: Gyeonggi.

KIWI_DIALECT_GYEONGSANG

Dialect flag: Gyeongsang.

KIWI_DIALECT_HAMGYEONG

Dialect flag: Hamgyeong.

KIWI_DIALECT_HWANGHAE

Dialect flag: Hwanghae.

KIWI_DIALECT_JEJU

Dialect flag: Jeju.

KIWI_DIALECT_JEOLLA

Dialect flag: Jeolla.

KIWI_DIALECT_PYEONGAN

Dialect flag: Pyeongan.

KIWI_DIALECT_STANDARD

Dialect mask: standard language only.

KIWI_MATCH_ALL

Common default match options.

KIWI_MATCH_ALL_WITH_NORMALIZING

KIWI_MATCH_ALL with coda normalization.

KIWI_MATCH_COMPATIBLE_JAMO

Match option: emit compatible jamo.

KIWI_MATCH_EMAIL

Match option: email detection.

KIWI_MATCH_HASHTAG

Match option: hashtag detection.

KIWI_MATCH_JOIN_ADJ_SUFFIX

Match option: join adjective suffixes.

KIWI_MATCH_JOIN_ADV_SUFFIX

Match option: join adverb suffixes.

KIWI_MATCH_JOIN_AFFIX

Match option convenience mask for all affix-join flags.

KIWI_MATCH_JOIN_NOUN_PREFIX

Match option: join noun prefixes.

KIWI_MATCH_JOIN_NOUN_SUFFIX

Match option: join noun suffixes.

KIWI_MATCH_JOIN_VERB_SUFFIX

Match option: join verb suffixes.

KIWI_MATCH_JOIN_V_SUFFIX

Match option convenience mask for verb/adjective suffix joins.

KIWI_MATCH_MENTION

Match option: mention detection.

KIWI_MATCH_MERGE_SAISIOT

Match option: merge saisiot.

KIWI_MATCH_NORMALIZE_CODA

Match option: normalize coda.

KIWI_MATCH_SERIAL

Match option: serial number detection.

KIWI_MATCH_SPLIT_COMPLEX

Match option: split complex morphemes.

KIWI_MATCH_SPLIT_SAISIOT

Match option: split saisiot.

KIWI_MATCH_URL

Match option: URL detection.

KIWI_MATCH_Z_CODA

Match option: z-coda handling.

KIWI_NUM_THREADS

Option key for setting worker threads through set_option.

KIWI_TYPO_BASIC_TYPO_SET

Typo preset: basic typo set.

KIWI_TYPO_BASIC_TYPO_SET_WITH_CONTINUAL

Typo preset: basic + continual typo sets.

KIWI_TYPO_BASIC_TYPO_SET_WITH_CONTINUAL_AND_LENGTHENING

Typo preset: basic + continual + lengthening typo sets.

KIWI_TYPO_CONTINUAL_TYPO_SET

Typo preset: continual typo set.

KIWI_TYPO_LENGTHENING_TYPO_SET

Typo preset: lengthening typo set.

KIWI_TYPO_WITHOUT_TYPO

Typo preset: disable typo correction.

Type Aliases

Analysis

Alias kept for readability in user code.

Result

Convenience alias used throughout the crate.