Function test_pypdfium2_parser

tests/main/test_pdf_parser.py:121–155

Dedicated functional test for the default `pypdfium2` PDF parser. `pypdfium2` is installed by default (core dependency), so this test does NOT require any optional `extras` to be installed -- it exercises the out-of-the-box PDF-parsing path that a bare `pip install langroid` gets.

(source: str)

@pytest.mark.parametrize("source", ["path", "bytes"])
def test_pypdfium2_parser(source: str):
    """
    Dedicated functional test for the default `pypdfium2` PDF parser.

    `pypdfium2` is installed by default (core dependency), so this test does
    NOT require any optional `extras` to be installed -- it exercises the
    out-of-the-box PDF-parsing path that a bare `pip install langroid` gets.
    """
    from langroid.parsing.document_parser import PyPDFium2Parser

    path = "tests/main/data/dummy.pdf"
    parser = DocumentParser.create(
        path, ParsingConfig(pdf=PdfParsingConfig(library="pypdfium2"))
    )
    assert isinstance(parser, PyPDFium2Parser)

    if source == "bytes":
        with open(path, "rb") as f:
            data = f.read()
        parser = DocumentParser.create(data, parser.config)
        assert isinstance(parser, PyPDFium2Parser)

    citation = path if source == "path" else "bytes"

    doc = parser.get_doc()
    assert isinstance(doc.content, str)
    # content correctness: known text from the sample PDF
    assert "Design and Evaluation" in doc.content
    assert "arXiv:2004.07606v1" in doc.content
    assert doc.metadata.source == citation

    chunks = parser.get_doc_chunks()
    assert len(chunks) > 0
    assert all(c.metadata.is_chunk for c in chunks)
    assert all(citation in c.metadata.source for c in chunks)


# @pytest.mark.skipif(