test_html_splitter_with_nested_preserved_elements() — langchain Function Reference

Source Code

libs/text-splitters/tests/unit_tests/test_text_splitters.py lines 3479–3516

def test_html_splitter_with_nested_preserved_elements() -> None:
    """Test HTML splitter with preserved elements nested in containers.

    Test that preserved elements are correctly preserved even when they are
    nested inside other container elements like <section> or <article>.
    This is a regression test for issue #31569
    """
    html_content = """
    <article>
        <h1>Section 1</h1>
        <section>
            <p>Some context about the data:</p>
            <table>
                <tr><td>Col1</td><td>Col2</td></tr>
                <tr><td>Data1</td><td>Data2</td></tr>
            </table>
            <p>Conclusion about data.</p>
        </section>
    </article>
    """
    with suppress_langchain_beta_warning():
        splitter = HTMLSemanticPreservingSplitter(
            headers_to_split_on=[("h1", "Header 1")],
            elements_to_preserve=["table"],
            max_chunk_size=1000,
        )
    documents = splitter.split_text(html_content)

    # The table should be preserved in the output
    assert len(documents) == 1
    content = documents[0].page_content
    # Check that the table structure is maintained (not flattened)
    assert "Col1" in content
    assert "Col2" in content
    assert "Data1" in content
    assert "Data2" in content
    # Check metadata
    assert documents[0].metadata == {"Header 1": "Section 1"}