mirror of
https://github.com/explosion/spaCy.git
synced 2024-12-28 19:06:33 +03:00
241d19a3e6
- added trailing space flags for tokens
31 lines
621 B
Python
31 lines
621 B
Python
# encoding: utf8
|
||
from __future__ import unicode_literals
|
||
|
||
from ..symbols import *
|
||
|
||
|
||
TOKENIZER_EXCEPTIONS = {
|
||
"Пн.": [
|
||
{ORTH: "Пн.", LEMMA: "Понедельник"}
|
||
],
|
||
"Вт.": [
|
||
{ORTH: "Вт.", LEMMA: "Вторник"}
|
||
],
|
||
"Ср.": [
|
||
{ORTH: "Ср.", LEMMA: "Среда"}
|
||
],
|
||
"Чт.": [
|
||
{ORTH: "Чт.", LEMMA: "Четверг"}
|
||
],
|
||
"Пт.": [
|
||
{ORTH: "Пт.", LEMMA: "Пятница"}
|
||
],
|
||
"Сб.": [
|
||
{ORTH: "Сб.", LEMMA: "Суббота"}
|
||
],
|
||
"Вс.": [
|
||
{ORTH: "Вс.", LEMMA: "Воскресенье"}
|
||
],
|
||
}
|
||
|