ai-content-maker/.venv/Lib/site-packages/spacy/tests/lang/ar/test_exceptions.py

21 lines
621 B
Python
Raw Normal View History

2024-05-03 04:18:51 +03:00
import pytest
@pytest.mark.parametrize("text", ["ق.م", "إلخ", "ص.ب", "ت."])
def test_ar_tokenizer_handles_abbr(ar_tokenizer, text):
tokens = ar_tokenizer(text)
assert len(tokens) == 1
def test_ar_tokenizer_handles_exc_in_text(ar_tokenizer):
text = "تعود الكتابة الهيروغليفية إلى سنة 3200 ق.م"
tokens = ar_tokenizer(text)
assert len(tokens) == 7
assert tokens[6].text == "ق.م"
def test_ar_tokenizer_handles_exc_in_text_2(ar_tokenizer):
text = "يبلغ طول مضيق طارق 14كم "
tokens = ar_tokenizer(text)
assert len(tokens) == 6