spaCy/spacy/tests/regression/test_issue351.py
2017-01-04 00:47:11 +01:00

17 lines
327 B
Python

from __future__ import unicode_literals
from ...en import English
import pytest
@pytest.fixture
def en_tokenizer():
return English.Defaults.create_tokenizer()
def test_issue351(en_tokenizer):
doc = en_tokenizer(" This is a cat.")
assert doc[0].idx == 0
assert len(doc[0]) == 3
assert doc[1].idx == 3