|
| 1 | +import os |
| 2 | +from unittest.mock import Mock, patch |
| 3 | + |
| 4 | +from langchain_core.documents import Document |
| 5 | + |
| 6 | +from template_langgraph.utilities.pdf_loaders import ( |
| 7 | + PdfLoaderWrapper, |
| 8 | + Settings, |
| 9 | + get_pdf_loader_settings, |
| 10 | +) |
| 11 | + |
| 12 | + |
| 13 | +class TestSettings: |
| 14 | + def test_default_settings(self): |
| 15 | + """Test default settings values.""" |
| 16 | + settings = Settings() |
| 17 | + assert settings.pdf_loader_data_dir_path == "./data" |
| 18 | + |
| 19 | + def test_custom_settings(self): |
| 20 | + """Test custom settings values.""" |
| 21 | + settings = Settings(pdf_loader_data_dir_path="/custom/path") |
| 22 | + assert settings.pdf_loader_data_dir_path == "/custom/path" |
| 23 | + |
| 24 | + @patch.dict(os.environ, {"PDF_LOADER_DATA_DIR_PATH": "/env/path"}) |
| 25 | + def test_env_settings(self): |
| 26 | + """Test settings from environment variables.""" |
| 27 | + settings = Settings() |
| 28 | + assert settings.pdf_loader_data_dir_path == "/env/path" |
| 29 | + |
| 30 | + |
| 31 | +class TestGetPdfLoaderSettings: |
| 32 | + def test_get_pdf_loader_settings_returns_settings(self): |
| 33 | + """Test that get_pdf_loader_settings returns a Settings instance.""" |
| 34 | + settings = get_pdf_loader_settings() |
| 35 | + assert isinstance(settings, Settings) |
| 36 | + |
| 37 | + def test_get_pdf_loader_settings_is_cached(self): |
| 38 | + """Test that get_pdf_loader_settings is cached.""" |
| 39 | + settings1 = get_pdf_loader_settings() |
| 40 | + settings2 = get_pdf_loader_settings() |
| 41 | + assert settings1 is settings2 |
| 42 | + |
| 43 | + |
| 44 | +class TestPdfLoaderWrapper: |
| 45 | + def test_init_with_default_settings(self): |
| 46 | + """Test PdfLoaderWrapper initialization with default settings.""" |
| 47 | + wrapper = PdfLoaderWrapper() |
| 48 | + assert isinstance(wrapper.settings, Settings) |
| 49 | + assert wrapper.settings.pdf_loader_data_dir_path == "./data" |
| 50 | + |
| 51 | + def test_init_with_custom_settings(self): |
| 52 | + """Test PdfLoaderWrapper initialization with custom settings.""" |
| 53 | + custom_settings = Settings(pdf_loader_data_dir_path="/custom/path") |
| 54 | + wrapper = PdfLoaderWrapper(settings=custom_settings) |
| 55 | + assert wrapper.settings.pdf_loader_data_dir_path == "/custom/path" |
| 56 | + |
| 57 | + @patch("template_langgraph.utilities.pdf_loaders.glob") |
| 58 | + @patch("template_langgraph.utilities.pdf_loaders.PyPDFLoader") |
| 59 | + def test_load_pdf_docs_no_files(self, mock_pdf_loader, mock_glob): |
| 60 | + """Test load_pdf_docs when no PDF files are found.""" |
| 61 | + mock_glob.return_value = [] |
| 62 | + wrapper = PdfLoaderWrapper() |
| 63 | + |
| 64 | + docs = wrapper.load_pdf_docs() |
| 65 | + |
| 66 | + assert docs == [] |
| 67 | + mock_glob.assert_called_once_with( |
| 68 | + os.path.join("./data", "**", "*.pdf"), |
| 69 | + recursive=True, |
| 70 | + ) |
| 71 | + |
| 72 | + @patch("template_langgraph.utilities.pdf_loaders.glob") |
| 73 | + @patch("template_langgraph.utilities.pdf_loaders.PyPDFLoader") |
| 74 | + def test_load_pdf_docs_with_files(self, mock_pdf_loader, mock_glob): |
| 75 | + """Test load_pdf_docs when PDF files are found.""" |
| 76 | + # Setup mock data |
| 77 | + mock_glob.return_value = ["./data/file1.pdf", "./data/file2.pdf"] |
| 78 | + |
| 79 | + mock_doc1 = Document(page_content="Content 1", metadata={"source": "file1.pdf"}) |
| 80 | + mock_doc2 = Document(page_content="Content 2", metadata={"source": "file1.pdf"}) |
| 81 | + mock_doc3 = Document(page_content="Content 3", metadata={"source": "file2.pdf"}) |
| 82 | + |
| 83 | + mock_loader_instance1 = Mock() |
| 84 | + mock_loader_instance1.load_and_split.return_value = [mock_doc1, mock_doc2] |
| 85 | + |
| 86 | + mock_loader_instance2 = Mock() |
| 87 | + mock_loader_instance2.load_and_split.return_value = [mock_doc3] |
| 88 | + |
| 89 | + mock_pdf_loader.side_effect = [mock_loader_instance1, mock_loader_instance2] |
| 90 | + |
| 91 | + wrapper = PdfLoaderWrapper() |
| 92 | + docs = wrapper.load_pdf_docs() |
| 93 | + |
| 94 | + # Verify results |
| 95 | + assert len(docs) == 3 |
| 96 | + assert docs[0] == mock_doc1 |
| 97 | + assert docs[1] == mock_doc2 |
| 98 | + assert docs[2] == mock_doc3 |
| 99 | + |
| 100 | + # Verify PyPDFLoader was called correctly |
| 101 | + assert mock_pdf_loader.call_count == 2 |
| 102 | + mock_pdf_loader.assert_any_call("./data/file1.pdf") |
| 103 | + mock_pdf_loader.assert_any_call("./data/file2.pdf") |
| 104 | + |
| 105 | + # Verify load_and_split was called with text splitter |
| 106 | + mock_loader_instance1.load_and_split.assert_called_once() |
| 107 | + mock_loader_instance2.load_and_split.assert_called_once() |
| 108 | + |
| 109 | + @patch("template_langgraph.utilities.pdf_loaders.glob") |
| 110 | + @patch("template_langgraph.utilities.pdf_loaders.PyPDFLoader") |
| 111 | + def test_load_pdf_docs_with_custom_data_dir(self, mock_pdf_loader, mock_glob): |
| 112 | + """Test load_pdf_docs with custom data directory.""" |
| 113 | + custom_settings = Settings(pdf_loader_data_dir_path="/custom/data") |
| 114 | + wrapper = PdfLoaderWrapper(settings=custom_settings) |
| 115 | + mock_glob.return_value = [] |
| 116 | + |
| 117 | + wrapper.load_pdf_docs() |
| 118 | + |
| 119 | + mock_glob.assert_called_once_with( |
| 120 | + os.path.join("/custom/data", "**", "*.pdf"), |
| 121 | + recursive=True, |
| 122 | + ) |
| 123 | + |
| 124 | + @patch("template_langgraph.utilities.pdf_loaders.glob") |
| 125 | + @patch("template_langgraph.utilities.pdf_loaders.PyPDFLoader") |
| 126 | + def test_load_pdf_docs_text_splitter_configuration(self, mock_pdf_loader, mock_glob): |
| 127 | + """Test that text splitter is configured correctly.""" |
| 128 | + mock_glob.return_value = ["./data/test.pdf"] |
| 129 | + mock_loader_instance = Mock() |
| 130 | + mock_loader_instance.load_and_split.return_value = [] |
| 131 | + mock_pdf_loader.return_value = mock_loader_instance |
| 132 | + |
| 133 | + wrapper = PdfLoaderWrapper() |
| 134 | + wrapper.load_pdf_docs() |
| 135 | + |
| 136 | + # Verify that load_and_split was called with a text splitter |
| 137 | + mock_loader_instance.load_and_split.assert_called_once() |
| 138 | + args = mock_loader_instance.load_and_split.call_args[0] |
| 139 | + assert len(args) == 1 |
| 140 | + text_splitter = args[0] |
| 141 | + |
| 142 | + # Verify text splitter configuration |
| 143 | + assert hasattr(text_splitter, "_chunk_size") |
| 144 | + assert hasattr(text_splitter, "_chunk_overlap") |
0 commit comments