|
| 1 | +import os |
| 2 | + |
| 3 | +from ckanext.spatial.harvesters.waf import _extract_waf |
| 4 | + |
| 5 | +TEST_DIR = os.path.dirname(os.path.abspath(__file__)) |
| 6 | +HTML_DIR = os.path.join(TEST_DIR, "html_files") |
| 7 | + |
| 8 | +def test_extract_iis(httpserver): |
| 9 | + |
| 10 | + # feed http response with these static html content |
| 11 | + with \ |
| 12 | + open(f"{HTML_DIR}/iis-folder.html", "r") as iis_folder, \ |
| 13 | + open(f"{HTML_DIR}/nginx-folder.html", "r") as nginx_folder, \ |
| 14 | + open(f"{HTML_DIR}/apache-folder.html", "r") as apache_folder, \ |
| 15 | + open(f"{HTML_DIR}/iis-subfolder.html", "r") as iis_subfolder, \ |
| 16 | + open(f"{HTML_DIR}/nginx-subfolder.html", "r") as nginx_subfolder, \ |
| 17 | + open(f"{HTML_DIR}/apache-subfolder.html", "r") as apache_subfolder: |
| 18 | + iis_folder_content = iis_folder.read() |
| 19 | + nginx_folder_content = nginx_folder.read() |
| 20 | + apache_folder_content = apache_folder.read() |
| 21 | + iis_subfolder_content = iis_subfolder.read() |
| 22 | + nginx_subfolder_content = nginx_subfolder.read() |
| 23 | + apache_subfolder_content = apache_subfolder.read() |
| 24 | + |
| 25 | + # feed static content when it traverses the subfolder |
| 26 | + httpserver.expect_request("/iis-folder/subfolder/").respond_with_data(iis_subfolder_content) |
| 27 | + httpserver.expect_request("/nginx-folder/subfolder/").respond_with_data(nginx_subfolder_content) |
| 28 | + httpserver.expect_request("/apache-folder/subfolder/").respond_with_data(apache_subfolder_content) |
| 29 | + |
| 30 | + # let it scape, traverse and extract the content |
| 31 | + iis_results = _extract_waf( |
| 32 | + iis_folder_content, |
| 33 | + httpserver.url_for("/iis-folder/"), |
| 34 | + "iis" |
| 35 | + ) |
| 36 | + |
| 37 | + nginx_results = _extract_waf( |
| 38 | + nginx_folder_content, |
| 39 | + httpserver.url_for("/nginx-folder/"), |
| 40 | + "nginx" |
| 41 | + ) |
| 42 | + |
| 43 | + apache_results = _extract_waf( |
| 44 | + apache_folder_content, |
| 45 | + httpserver.url_for("/apache-folder/"), |
| 46 | + "apache" |
| 47 | + ) |
| 48 | + |
| 49 | + records_expected = [('record-1.xml', '2024-11-07 15:00:00'), ('record-2.xml', '2024-11-07 16:59:00')] |
| 50 | + |
| 51 | + assert records_expected == sorted([(os.path.basename(r[0]), r[1]) for r in iis_results]) |
| 52 | + assert records_expected == sorted([(os.path.basename(r[0]), r[1]) for r in nginx_results]) |
| 53 | + assert records_expected == sorted([(os.path.basename(r[0]), r[1]) for r in apache_results]) |
0 commit comments