mirror of
https://github.com/explosion/spaCy.git
synced 2024-12-27 02:16:32 +03:00
15 lines
323 B
Python
15 lines
323 B
Python
from __future__ import unicode_literals
|
|
|
|
import pytest
|
|
|
|
def test_hyphen(en_tokenizer):
|
|
tokens = en_tokenizer('best-known')
|
|
assert len(tokens) == 3
|
|
|
|
|
|
def test_period(en_tokenizer):
|
|
tokens = en_tokenizer('best.Known')
|
|
assert len(tokens) == 3
|
|
tokens = en_tokenizer('zombo.com')
|
|
assert len(tokens) == 1
|