spaCy/tests/test_urlish.py

21 lines
483 B
Python
Raw Normal View History

2014-11-02 13:23:04 +00:00
from spacy.orth import like_url
2014-11-02 02:21:57 +00:00
def test_basic_url():
2014-11-02 13:23:04 +00:00
assert like_url('www.google.com')
assert like_url('google.com')
assert like_url('sydney.com')
assert like_url('Sydney.edu')
assert like_url('2girls1cup.org')
2014-11-02 02:21:57 +00:00
def test_close_enough():
2014-11-02 13:23:04 +00:00
assert like_url('http://stupid')
assert like_url('www.hi')
2014-11-02 02:21:57 +00:00
def test_non_match():
2014-11-02 13:23:04 +00:00
assert not like_url('dog')
assert not like_url('1.2')
assert not like_url('1.a')
assert not like_url('hello.There')