spaCy/tests/parser/test_sbd.py
2015-06-07 18:02:24 +02:00

36 lines
948 B
Python

from __future__ import unicode_literals
import pytest
def test_single_period(EN):
string = 'A test sentence.'
words = EN(string)
assert len(words) == 4
assert len(list(words.sents)) == 1
assert sum(len(sent) for sent in words.sents) == len(words)
def test_single_no_period(EN):
string = 'A test sentence'
words = EN(string)
assert len(words) == 3
assert len(list(words.sents)) == 1
assert sum(len(sent) for sent in words.sents) == len(words)
def test_single_exclamation(EN):
string = 'A test sentence!'
words = EN(string)
assert len(words) == 4
assert len(list(words.sents)) == 1
assert sum(len(sent) for sent in words.sents) == len(words)
def test_single_question(EN):
string = 'A test sentence?'
words = EN(string, tag=False, parse=False)
assert len(words) == 4
assert len(list(words.sents)) == 1
assert sum(len(sent) for sent in words.sents) == len(words)