5d4a480c48a242c1d8051be502ecb98a9b273e9c,tests/candidates/test_candidates.py,,test_ngram_split,#Any#,33

Before Change

    caplog.set_level(logging.INFO)
    PARALLEL = 1
    max_docs = 1
    session = Meta.init("postgres://localhost:5432/" + DB).Session()

    docs_path = "tests/data/html/112823.html"
    pdf_path = "tests/data/pdf/112823.pdf"

After Change

    // When a text starts with a split_token.
    sent.text = "-Text"
    sent.words = ["-Text"]
    result = list(ngrams.apply(sent))

    assert len(result) == 2
    assert result[0].get_span() == "-Text"

Instances