Function test_document_chunking

py/tests/unit/document/test_document_processing.py:54–95

Test document chunking functionality.

(mock_document_handler, sample_document)

@pytest.mark.asyncio
async def test_document_chunking(mock_document_handler, sample_document):
    """Test document chunking functionality."""
    from core.main.services.documents import DocumentProcessingService

    # Setup the chunking service with mocked components
    service = DocumentProcessingService(document_handler=mock_document_handler)

    # Mock the chunking method
    original_chunk_method = service.chunk_document
    service.chunk_document = MagicMock(return_value=[
        DocumentChunk(
            chunk_id="new-chunk-1",
            document_id=sample_document.document_id,
            text="Aristotle was a Greek philosopher.",
            metadata={"auto_chunk": True}
        ),
        DocumentChunk(
            chunk_id="new-chunk-2",
            document_id=sample_document.document_id,
            text="He studied under Plato.",
            metadata={"auto_chunk": True}
        ),
        DocumentChunk(
            chunk_id="new-chunk-3",
            document_id=sample_document.document_id,
            text="He made significant contributions to logic, ethics, and metaphysics.",
            metadata={"auto_chunk": True}
        )
    ])

    # Process the document
    processed_doc = await service.process_document(sample_document)

    # Verify chunking was called
    service.chunk_document.assert_called_once()

    # Check that document was updated with new chunks
    assert len(processed_doc.chunks) == 3
    assert all(chunk.metadata.get("auto_chunk") for chunk in processed_doc.chunks)

    # Restore original method
    service.chunk_document = original_chunk_method


@pytest.mark.asyncio