Language identification based on fasttext (lid.176.ftz https://fasttext.cc/docs/en/language-identification.html).
The lid.176.ftz
file is licensed under Creative Commons Attribution-Share-Alike License 3.0 and is not part of this module. It is automatically downloaded from its external origin on the first run of this module.
This module attempts to immitate the follow two features of langid
- langid.classify: fastlid
- langid.set_languages(langs=[...]): fastlid.set_languages = [...]
- import fastlid
- fastlid.set_languages = ['nl','fr'])
- TODO: Commandline interface
pip install fasttext
For Windows without a C/C++ compiler:
- Download a proper whl (e.g.,
fasttext‑0.9.2‑cp36‑cp36m‑win_amd64.whl
for 64bits Python 3.6 etc) from https://www.lfd.uci.edu/~gohlke/pythonlibs/#fasttext
pip install fasttext*.whl
or (for python 3.8)
pip install https://github.com/ffreemt/ezbee/raw/main/data/artifects/fasttext-0.9.2-cp38-cp38-win_amd64.whl
pip install fastlid
or install from git
pip install git+https://github.com/ffreemt/fast-langid.git
# also works pip install git+https://github.com/ffreemt/fast-langid
or clone the git repo and install from source.
from fastlid import fastlid, supported_langs
# support 176 languages
print(supported_langs, len(supported_langs))
# ['af', 'als', 'am', 'an', 'ar', 'arz', 'as', 'ast', 'av', 'az'] 176
fastlid("test this")
# ('en', 0.765)
fastlid("test this 测试一下", k=2)
# (['zh', 'en'], [0.663, 0.124])
fastlid.set_languages = ['fr', 'zh']
fastlid("test this 测试吧")
# ('zh', 0.01)
fastlid.set_languages = None
fastlid("test this 测试吧")
('en', 0.686)
fastlid.set_languages = ['fr', 'zh', 'en']
fastlid("test this 测试吧", k=3)
(['en', 'zh', 'fr'], [0.686, 0.01, 0.006])
N.B. hanzidentifier
can be used to identify simplified Chinese or/and traditional Chinese should you need to do so.
Install poetry
and yarn
the way you like it.
poetry install # install python packages
yarn install --dev # install necesary node packages
# ...code...
yarn test
yarn final
# ...optionally submit pr...