JamoLib

Fast Hangul text utilities for Python. JamoLib helps you decompose Hangul syllables into jamo, compose jamo back into syllables, and convert two-set Korean keyboard input into readable Korean text.

Why JamoLib

• Linear-time text scanning for large Hangul strings.
• Small API surface that is easy to drop into search, normalization, keyboard-input, and NLP preprocessing pipelines.
• Preserves mixed text such as English, numbers, punctuation, and whitespace.
• Handles common batchim boundary cases like 값이, 닭이, and 읽어.
• Supports common compound-medial combinations such as ㄱㅗㅏ -> 과.

Installation

pip install jamolib

Quick Start

import jamolib

text = "한글과 English 123"
decomposed = jamolib.decomposeHangulText(text)

print(decomposed)
# ㅎㅏㄴㄱㅡㄹㄱㅘ English 123

print(jamolib.composeHangulText("ㄱㅏㅂㅅㅇㅣ"))
# 값이

print(jamolib.translateEngToKor("dkssudgktpdy"))
# 안녕하세요

API At A Glance

Examples

• examples/quickstart.py: decomposition, composition, and keyboard conversion in one script
• examples/mixed_text.py: preserving non-Hangul text while processing Hangul
• examples/batchim_boundaries.py: tricky batchim and syllable-boundary cases

Run an example from the repository root:

python examples/quickstart.py

Notes

• decomposeHangul expects a single Hangul syllable.
• composeHangul expects compatibility jamo in the order 초성 + 중성 [+ 종성].
• composeHangulText also combines common compound medials like ㅗㅏ, ㅜㅓ, and ㅡㅣ.
• translateEngToKor uses the standard two-set Korean keyboard mapping.
• Mixed strings are preserved as-is outside Hangul processing.

Performance

The current implementation uses a single-pass scanner instead of repeated global string replacement. Local measurements on Python 3.12 in this repository produced the following averages:

These numbers are environment-dependent, but they reflect the optimized code currently in the repository.

Development

python -m pip install -e .[test]
pytest
python scripts/benchmark.py
python -m build

Project Docs

• Changelog
• Contributing guide
• Issue tracker
• Releases