rhoknp
raw JSON → 1.8.0 verified Fri May 01 auth: no python
Yet another Python binding for Juman++/KNP/KWJA. Current version: 1.8.0. Provides a modern Python interface for Japanese morphological analysis (Juman++), dependency/case structure analysis (KNP), and anaphora resolution (KWJA). Follows semantic versioning with frequent patch releases.
pip install rhoknp Common errors
error ModuleNotFoundError: No module named 'rhoknp.juman' ↓
cause Importing from deprecated submodule path used in versions <1.0.0.
fix
Change import to 'from rhoknp import Jumanpp'.
error JumanppNotFoundError: Juman++ command not found ↓
cause The jumanpp executable is not installed or not on PATH.
fix
Install Juman++ (e.g., 'apt install jumanpp' on Ubuntu or download from https://github.com/ku-nlp/jumanpp).
Warnings
breaking In v1.0.0, the import path changed from 'rhoknp.juman.jumanpp' to 'rhoknp'. Old syntax no longer works. ↓
fix Use 'from rhoknp import Jumanpp' instead of 'from rhoknp.juman.jumanpp import Jumanpp'.
deprecated The 'analyze_sentence' method is deprecated in v1.5.0. Use 'analyze' instead. ↓
fix Replace jumanpp.analyze_sentence(text) with jumanpp.analyze(text).
gotcha The Jumanpp/Knp/Kwja objects require the underlying executables (jumanpp, knp, kwja) to be installed on the system PATH. pip install rhoknp does not install them. ↓
fix Install the required binaries separately (e.g., via apt, Homebrew, or source).
Imports
- Jumanpp
from rhoknp import Jumanpp - Knp
from rhoknp import Knp - Kwja
from rhoknp import Kwja - Document
from rhoknp import Document
Quickstart
from rhoknp import Jumanpp
jumanpp = Jumanpp()
result = jumanpp.analyze("太郎は本を読んだ。")
for morpheme in result.morphemes:
print(morpheme.surf, morpheme.reading)