spaCy/spacy/lang/zh/__init__.py
2018-02-25 15:12:38 +01:00

42 lines
1.3 KiB
Python

# coding: utf8
from __future__ import unicode_literals
from ...attrs import LANG
from ...language import Language
from ...tokens import Doc
class ChineseDefaults(Language.Defaults):
lex_attr_getters = dict(Language.Defaults.lex_attr_getters)
lex_attr_getters[LANG] = lambda text: 'zh' # for pickling
use_jieba = True
class Chinese(Language):
lang = 'zh'
Defaults = ChineseDefaults # override defaults
def make_doc(self, text):
if self.Defaults.use_jieba:
try:
import jieba
except ImportError:
msg = ("Jieba not installed. Either set Chinese.use_jieba = False, "
"or install it https://github.com/fxsjy/jieba")
raise ImportError(msg)
words = list(jieba.cut(text, cut_all=False))
words = [x for x in words if x]
return Doc(self.vocab, words=words, spaces=[False]*len(words))
else:
words = []
spaces = []
doc = self.tokenizer(text)
for token in self.tokenizer(text):
words.extend(list(token.text))
spaces.extend([False]*len(token.text))
spaces[-1] = bool(token.whitespace_)
return Doc(self.vocab, words=words, spaces=spaces)
__all__ = ['Chinese']