test: added expected structure to ensure correctness

Signed-off-by: Carl Flottmann <[email protected]>

Author: art1f1c3R

tests/malware_analyzer/pypi/test_similar_projects.py

@@ -82,9 +82,62 @@ def test_get_normalized_structure(
     mock_send_get_http_raw: Callable
         A mocked send_get_http_raw function to return inspector html.
     """
+    # This structure is all files in the PyPI inspector HTML page that are python files, minus setup.py
+    # and the package name, so this structure is now comparable to some other project that would have these
+    # same files, but with a different name
+    expected_structure: set[str] = {
+        "/__init__.py",
+        "/ai/__init__.py",
+        "/ai/anthropic/__init__.py",
+        "/ai/anthropic/anthropic.py",
+        "/ai/anthropic/anthropic_async.py",
+        "/ai/anthropic/anthropic_converter.py",
+        "/ai/anthropic/anthropic_providers.py",
+        "/ai/gemini/__init__.py",
+        "/ai/gemini/gemini.py",
+        "/ai/gemini/gemini_converter.py",
+        "/ai/langchain/__init__.py",
+        "/ai/langchain/callbacks.py",
+        "/ai/openai/__init__.py",
+        "/ai/openai/openai.py",
+        "/ai/openai/openai_async.py",
+        "/ai/openai/openai_converter.py",
+        "/ai/openai/openai_providers.py",
+        "/ai/sanitization.py",
+        "/ai/types.py",
+        "/ai/utils.py",
+        "/args.py",
+        "/client.py",
+        "/consumer.py",
+        "/contexts.py",
+        "/exception_capture.py",
+        "/exception_utils.py",
+        "/feature_flags.py",
+        "/integrations/__init__.py",
+        "/integrations/django.py",
+        "/poller.py",
+        "/request.py",
+        "/test/__init__.py",
+        "/test/test_before_send.py",
+        "/test/test_client.py",
+        "/test/test_consumer.py",
+        "/test/test_contexts.py",
+        "/test/test_exception_capture.py",
+        "/test/test_feature_flag.py",
+        "/test/test_feature_flag_result.py",
+        "/test/test_feature_flags.py",
+        "/test/test_module.py",
+        "/test/test_request.py",
+        "/test/test_size_limited_dict.py",
+        "/test/test_types.py",
+        "/test/test_utils.py",
+        "/types.py",
+        "/utils.py",
+        "/version.py",
+    }
+
     pypi_package_json.component_version = "6.7.4"
     pypi_package_json.component_name = "posthog"
-
     pypi_package_json.inspector_asset.package_sdist_link = TEST_SDIST_LINK
     pypi_package_json.inspector_asset.package_link_reachability = {TEST_SDIST_LINK: True}
     pypi_package_json.inspector_asset.package_whl_links = []
@@ -96,3 +149,6 @@ def test_get_normalized_structure(
     wheel_structure = analyzer.get_normalized_structure(pypi_package_json)
 
     assert sdist_structure == wheel_structure
+    assert sdist_structure == expected_structure
+    # sdist == wheel_structure, sdist == expected_structure, so don't quite need this, but just for completeness
+    assert wheel_structure == expected_structure