[MNT]Replaced Live Test-Server Calls with Mocks in Unit Tests on tests/test_study/test_study_function.py module

[Sachin0496]

Metadata

• Reference Issue: Fixes [MNT] Replace Live Server Calls with Mocks in Unit Tests #1649
• New Tests Added: No
• Documentation Updated: No
• Change Log Entry: Refactored study tests by introducing a centralized StudyMockServer that mocks requests.Session calls, replacing live test-server interactions. This reduced the request load on test_server and production_server, improved maintainability, and reduced average test runtime from ~30s to ~8s.

---

Details

What does this PR implement?

This PR completely overhauls the mocking strategy in tests/test_study/test_study_functions.py.

Previously, the tests relied heavily on live hits to the OpenML test server and a massive amount of cluttered, inline XML string generation. This PR extracts all of that logic into a dedicated, standalone StudyMockServer helper class that dynamically generates the necessary mocked API responses.

Why is this change necessary?

The existing study tests had two major problems:

1. Performance & Reliability: Relying on the live OpenML test server meant tests were slow, flaky, and put unnecessary load on the server.

Impact of this PR:

• Massive Speedup: By properly mocking these server interactions, the average test execution time for this suite dropped from ~30s to ~8s.
• Reduced Server Load: Tests have been mocked/patched with @mock.patch.object(requests.Session), reducing the load on live server

---

Implementation Overview

Mock Strategy

All study test-server interactions (GET, POST, DELETE) are intercepted using @mock.patch.object(requests.Session). However, the setup is now entirely delegated:

1. StudyMockServer Class: A new static helper class that encapsulates all XML generation methods (build_study_xml, build_runs_xml, etc.).
2. Dedicated Mock Setups: Extracted massive side_effect assignments into descriptive static methods (e.g., StudyMockServer.setup_publish_study_mocks).
3. run_ids Support: Fixed a bug in the old XML generation by updating build_study_xml to accept an explicit list of run_ids. This ensures the mock server returns the exact IDs the test expects after an attach/detach operation without failing assertions.

---

Refactored Test Cases

The following test methods were dramatically simplified by offloading their mock setups to the StudyMockServer:

• test_publish_benchmark_suite: Mocks centralized via StudyMockServer.setup_publish_benchmark_suite_mocks.
• test_publish_study: Mocks centralized via StudyMockServer.setup_publish_study_mocks. Fixed the bug causing assertion failures on detached run IDs.
• test_publish_empty_study_explicit & implicit: Cleaned up copy-paste errors and centralized the empty XML mock generation.
• test_study_attach_illegal: Mocks centralized via StudyMockServer.setup_study_attach_illegal_mocks, properly injecting openml.exceptions.OpenMLServerException to simulate server-side validation errors.

---

What Was NOT Changed

• Test Logic: The actual test execution, API calls being tested, and assertions remain 100% identical.
• Production Server Tests: Tests hitting the real server (e.g., test_get_study_old, test_get_suite) remain untouched to continue validating real server behavior.
• Marks and skips: The server markers such as pytest.mark.test_server and skips such as @unittest.skip were left untouched , still mocks have been implemented for the same

---

How to Reproduce

Run the study test suite locally to verify the speedup and clean output. No OpenML test server access is required for these specific tests anymore:

pytest tests/test_study/test_study_functions.py