mirror of https://github.com/explosion/spaCy.git
42 lines
1.3 KiB
Python
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']
|