spaCy/spacy/pt/__init__.py
2017-01-08 22:17:37 +01:00

19 lines
451 B
Python

# encoding: utf8
from __future__ import unicode_literals, print_function
from ..language import Language
from ..attrs import LANG
from .language_data import *
class Portuguese(Language):
lang = 'pt'
class Defaults(Language.Defaults):
lex_attr_getters = dict(Language.Defaults.lex_attr_getters)
lex_attr_getters[LANG] = lambda text: 'pt'
tokenizer_exceptions = TOKENIZER_EXCEPTIONS
stop_words = STOP_WORDS