mirror of
https://github.com/explosion/spaCy.git
synced 2024-12-27 18:36:36 +03:00
c09d2fa25b
* luganda language extension * __init__.py changes * New enhancements * Lexical attribute changed * punctuaction and sentence additions * Remove comment header * Fix typos, reformat * reformated version * Add tokenizer test * Remove contractions from stop words * Format * Add Luganda to website Co-authored-by: Adriane Boyd <adrianeboyd@gmail.com>
16 lines
449 B
Python
16 lines
449 B
Python
import pytest
|
||
|
||
LG_BASIC_TOKENIZATION_TESTS = [
|
||
(
|
||
"Abooluganda ab’emmamba ababiri",
|
||
["Abooluganda", "ab’emmamba", "ababiri"],
|
||
),
|
||
]
|
||
|
||
|
||
@pytest.mark.parametrize("text,expected_tokens", LG_BASIC_TOKENIZATION_TESTS)
|
||
def test_lg_tokenizer_basic(lg_tokenizer, text, expected_tokens):
|
||
tokens = lg_tokenizer(text)
|
||
token_list = [token.text for token in tokens if not token.is_space]
|
||
assert expected_tokens == token_list
|