CoolFace
Apppublic

MegaTronX/Subsystem_OS_Command_Access

sourceHugging Faceupdated 2y agoView on Hugging Face
0likes
test_seg_text.py48 linesDownload Raw Back to tests
1"""Test seg_text."""2import pytest3from gradiobee.seg_text import seg_text4 5 6def test_seg_text1():7    """Test seg_text 1."""8    text = " text 1\n\n test 2. test 3"9    _ = seg_text(text)10    assert len(_) == 211 12    text = " text 1\n\n test 2. Test 3"13    _ = seg_text(text)14    assert len(_) == 315 16 17@pytest.mark.parametrize(18    "test_input,expected", [19        ("", []),20        (" ", []),21        (" \n ", []),22    ]23)24def test_seg_text_blanks(test_input, expected):25    """Test blanks."""26    assert seg_text(test_input) == expected27 28 29def test_seg_text_semicolon ():30    """Test semicolon."""31    text = """ “元宇宙”,英文為“Metaverse”。該詞出自1992年;的科幻小說《雪崩》。 """32    assert len(seg_text(text)) == 233    assert len(seg_text(text, 'zh')) == 234    assert len(seg_text(text, 'ja')) == 235    assert len(seg_text(text, 'ko')) == 236    assert len(seg_text(text, 'en')) == 137 38 39def test_seg_text_semicolon_extra ():40    """Test semicolon."""41    extra = "[;;]"42    text = """ “元宇宙”,英文為“Metaverse”。該詞出自1992年;的科幻小說《雪崩》。 """43    assert len(seg_text(text, extra=extra)) == 2 + 144    assert len(seg_text(text, 'zh', extra=extra)) == 2 + 145    assert len(seg_text(text, 'ja', extra=extra)) == 2 + 146    assert len(seg_text(text, 'ko', extra=extra)) == 2 + 147    assert len(seg_text(text, 'en', extra=extra)) == 1 + 148