2017-01-12 14:27:46 +00:00
|
|
|
import pytest
|
2021-06-02 09:16:57 +00:00
|
|
|
from spacy.attrs import LEMMA, ORTH, IS_ALPHA
|
2018-07-24 21:38:44 +00:00
|
|
|
from spacy.parts_of_speech import NOUN, VERB
|
2015-10-26 01:31:05 +00:00
|
|
|
|
|
|
|
|
2018-11-27 00:09:36 +00:00
|
|
|
@pytest.mark.parametrize(
|
|
|
|
"text1,text2", [("Hello", "bye"), ("Hello", "hello"), ("Hello", "Hello,")]
|
|
|
|
)
|
2017-01-12 14:27:46 +00:00
|
|
|
def test_vocab_api_neq(en_vocab, text1, text2):
|
|
|
|
assert en_vocab[text1].orth != en_vocab[text2].orth
|
2015-10-26 01:31:05 +00:00
|
|
|
|
|
|
|
|
2018-11-27 00:09:36 +00:00
|
|
|
@pytest.mark.parametrize("text", "Hello")
|
2017-01-12 14:27:46 +00:00
|
|
|
def test_vocab_api_eq(en_vocab, text):
|
|
|
|
lex = en_vocab[text]
|
|
|
|
assert en_vocab[text].orth == lex.orth
|
2015-10-26 01:31:05 +00:00
|
|
|
|
|
|
|
|
2018-11-27 00:09:36 +00:00
|
|
|
@pytest.mark.parametrize("text", ["example"])
|
2017-01-12 14:27:46 +00:00
|
|
|
def test_vocab_api_shape_attr(en_vocab, text):
|
|
|
|
lex = en_vocab[text]
|
|
|
|
assert lex.orth != lex.shape
|
2015-10-26 01:31:05 +00:00
|
|
|
|
|
|
|
|
2018-11-27 00:09:36 +00:00
|
|
|
@pytest.mark.parametrize(
|
|
|
|
"string,symbol",
|
|
|
|
[
|
|
|
|
("IS_ALPHA", IS_ALPHA),
|
|
|
|
("NOUN", NOUN),
|
|
|
|
("VERB", VERB),
|
|
|
|
("LEMMA", LEMMA),
|
|
|
|
("ORTH", ORTH),
|
|
|
|
],
|
|
|
|
)
|
2017-01-12 14:27:46 +00:00
|
|
|
def test_vocab_api_symbols(en_vocab, string, symbol):
|
|
|
|
assert en_vocab.strings[string] == symbol
|
2015-10-26 01:31:05 +00:00
|
|
|
|
|
|
|
|
2018-11-27 00:09:36 +00:00
|
|
|
@pytest.mark.parametrize("text", "Hello")
|
2017-01-13 13:26:53 +00:00
|
|
|
def test_vocab_api_contains(en_vocab, text):
|
2018-11-30 16:43:08 +00:00
|
|
|
_ = en_vocab[text] # noqa: F841
|
2017-01-12 14:27:46 +00:00
|
|
|
assert text in en_vocab
|
|
|
|
assert "LKsdjvlsakdvlaksdvlkasjdvljasdlkfvm" not in en_vocab
|
2019-03-11 14:23:20 +00:00
|
|
|
|
|
|
|
|
|
|
|
def test_vocab_writing_system(en_vocab):
|
|
|
|
assert en_vocab.writing_system["direction"] == "ltr"
|
2019-03-11 14:28:22 +00:00
|
|
|
assert en_vocab.writing_system["has_case"] is True
|