|
| 1 | +import os |
| 2 | +import pathlib |
| 3 | + |
| 4 | +import pytest |
| 5 | + |
| 6 | +from unstructured.ingest.connector.git import GitIngestDoc, SimpleGitConfig |
| 7 | + |
| 8 | +DIRECTORY = pathlib.Path(__file__).parent.resolve() |
| 9 | +EXAMPLE_DOCS_DIRECTORY = os.path.join(DIRECTORY, "..", "example-docs") |
| 10 | + |
| 11 | +test_files = [ |
| 12 | + "layout-parser-paper-fast.jpg", |
| 13 | + "layout-parser-paper-fast.pdf", |
| 14 | +] |
| 15 | + |
| 16 | + |
| 17 | +@pytest.mark.parametrize("filename", test_files) |
| 18 | +def test_process_file_include_filename(filename: str): |
| 19 | + ingest_doc = GitIngestDoc( |
| 20 | + path=filename, |
| 21 | + config=SimpleGitConfig( |
| 22 | + download_dir=EXAMPLE_DOCS_DIRECTORY, |
| 23 | + metadata_include="filename", |
| 24 | + ), |
| 25 | + ) |
| 26 | + isd_elems = ingest_doc.process_file() |
| 27 | + |
| 28 | + for elem in isd_elems: |
| 29 | + for k in elem["metadata"]: |
| 30 | + assert k == "filename" |
| 31 | + |
| 32 | + |
| 33 | +@pytest.mark.parametrize("filename", test_files) |
| 34 | +def test_process_file_include_filename_pagenum(filename: str): |
| 35 | + ingest_doc = GitIngestDoc( |
| 36 | + path=filename, |
| 37 | + config=SimpleGitConfig( |
| 38 | + download_dir=EXAMPLE_DOCS_DIRECTORY, |
| 39 | + metadata_include="filename,page_number", |
| 40 | + ), |
| 41 | + ) |
| 42 | + isd_elems = ingest_doc.process_file() |
| 43 | + |
| 44 | + for elem in isd_elems: |
| 45 | + for k in elem["metadata"]: |
| 46 | + assert k in ["filename", "page_number"] |
| 47 | + |
| 48 | + |
| 49 | +@pytest.mark.parametrize("filename", test_files) |
| 50 | +def test_process_file_exclude_filename(filename: str): |
| 51 | + ingest_doc = GitIngestDoc( |
| 52 | + path=filename, |
| 53 | + config=SimpleGitConfig( |
| 54 | + download_dir=EXAMPLE_DOCS_DIRECTORY, |
| 55 | + metadata_exclude="filename", |
| 56 | + ), |
| 57 | + ) |
| 58 | + isd_elems = ingest_doc.process_file() |
| 59 | + |
| 60 | + for elem in isd_elems: |
| 61 | + for k in elem["metadata"]: |
| 62 | + assert k != "filename" |
| 63 | + |
| 64 | + |
| 65 | +@pytest.mark.parametrize("filename", test_files) |
| 66 | +def test_process_file_exclude_filename_pagenum(filename: str): |
| 67 | + ingest_doc = GitIngestDoc( |
| 68 | + path=filename, |
| 69 | + config=SimpleGitConfig( |
| 70 | + download_dir=EXAMPLE_DOCS_DIRECTORY, |
| 71 | + metadata_exclude="filename,page_number", |
| 72 | + ), |
| 73 | + ) |
| 74 | + isd_elems = ingest_doc.process_file() |
| 75 | + |
| 76 | + for elem in isd_elems: |
| 77 | + for k in elem["metadata"]: |
| 78 | + assert k not in ["filename", "page_number"] |
0 commit comments