Enhance tests to validate fallback behavior for virus scan errors and ensure save_file is called in main flow

Author: dylanpicart

tests/test_excel_scraper.py

@@ -157,7 +157,7 @@ def mock_stream(method, _url, timeout=10):
 async def test_download_excel_scan_error(test_scraper):
     """
     Test the scenario where virus scanning fails (connection reset, etc.).
-    The method should also skip returning file content, but not label it as a virus.
+    If MIME is valid, content should still be returned (fallback behavior).
     """
     fake_excel_content = b"FakeExcelData"
     url = "http://example.com/scan_error.xls"
@@ -179,14 +179,15 @@ async def __aexit__(self, exc_type, exc_val, exc_tb):
     def mock_stream(method, _url, timeout=10):
         return MockResponseContext(200, fake_excel_content)
 
-    # "ERROR" indicates scanning couldn't complete
+    # "ERROR" means scanning failed, but MIME check passes
     with patch.object(test_scraper.session, 'stream', side_effect=mock_stream), \
-         patch.object(test_scraper._security_manager, 'scan_for_viruses', return_value=("ERROR","Connection reset by peer")), \
+         patch.object(test_scraper._security_manager, 'scan_for_viruses', return_value=("ERROR", "Connection reset")), \
          patch.object(test_scraper._security_manager, 'is_excel_file', return_value=True):
 
         returned_url, content = await test_scraper.download_excel(url)
         assert returned_url == url
-        assert content is None, "Should skip returning content if scan error occurs."
+        assert content == fake_excel_content  # ✅ New expected behavior
+
 
 @pytest.mark.asyncio
 async def test_download_excel_not_excel(test_scraper):

tests/test_main.py

@@ -40,16 +40,19 @@ async def test_main_scraper_flow():
     # IMPORTANT: Patch the methods on the actual module path where
     # `NYCInfoHubScraper` is defined and imported by main.py—i.e. "src.excel_scraper"
     with patch("src.excel_scraper.NYCInfoHubScraper.scrape_excel_links", return_value=mock_excel_links), \
-         patch("src.excel_scraper.NYCInfoHubScraper.concurrent_fetch", return_value=mock_files_map), \
-         patch("src.excel_scraper.NYCInfoHubScraper.parallel_hashing", return_value=mock_hashes), \
-         patch("src.excel_scraper.NYCInfoHubScraper.save_file") as mock_save:
+        patch("src.excel_scraper.NYCInfoHubScraper.concurrent_fetch", return_value=mock_files_map), \
+        patch("src.excel_scraper.NYCInfoHubScraper.parallel_hashing", return_value=mock_hashes), \
+        patch("src.excel_scraper.NYCInfoHubScraper.save_file") as mock_save:
 
         exit_code = await main_entrypoint()
 
-        # Verify main() completed successfully
+        # ✅ Check main ran successfully
         assert exit_code == 0, "Expected main to return 0 on success"
 
-        # Check calls to save_file
+        # ✅ NEW: Confirm the mock was even called
+        assert mock_save.called, "save_file was not called at all"
+
+        # ✅ Check expected calls
         mock_save.assert_any_call(
             "http://example.com/attendance_2021.xlsx",
             b"fake attendance bytes",
@@ -61,3 +64,4 @@ async def test_main_scraper_flow():
             "hash2"
         )
         assert mock_save.call_count == 2, "Expected two calls to save_file"
+