Spaces:
Sleeping
Sleeping
| """ | |
| Unit tests for the grading system. | |
| Run with: pytest tests/test_graders.py -v | |
| """ | |
| import sys | |
| sys.path.insert(0, ".") | |
| import pytest | |
| from tasks import TASKS | |
| from graders import grade_submission, normalize_rows, compute_episode_score | |
| def test_correct_solutions_score_1(): | |
| for task_id, task in TASKS.items(): | |
| score, feedback, _ = grade_submission(task_id, task["correct_query"]) | |
| assert score == 1.0, f"{task_id}: correct query scored {score} — {feedback}" | |
| def test_wrong_query_scores_below_1(): | |
| for task_id in list(TASKS.keys())[:3]: | |
| score, _, _ = grade_submission(task_id, "SELECT 'wrong'") | |
| assert score < 1.0, f"{task_id}: wrong query should score < 1.0, got {score}" | |
| def test_buggy_query_scores_below_1(): | |
| for task_id, task in TASKS.items(): | |
| score, _, _ = grade_submission(task_id, task["buggy_query"]) | |
| assert score < 1.0, f"{task_id}: buggy query should not score 1.0" | |
| def test_scores_in_range(): | |
| for task_id, task in TASKS.items(): | |
| for query in [task["correct_query"], task["buggy_query"], "SELECT 1"]: | |
| score, _, _ = grade_submission(task_id, query) | |
| assert 0.0 <= score <= 1.0, f"{task_id}: score {score} out of range" | |
| def test_deterministic(): | |
| query = "SELECT first_name FROM employees" | |
| scores = [grade_submission("easy_missing_from", query)[0] for _ in range(5)] | |
| assert len(set(scores)) == 1, f"Non-deterministic scores: {scores}" | |
| def test_normalize_rows(): | |
| rows = [(1, "Hello ", 3.14159), (2, "world", 2.71828)] | |
| normalized = normalize_rows(rows) | |
| assert normalized[0][2] == 3.14 | |
| assert normalized[0][1] == "Hello" | |
| def test_episode_score_early_solve_bonus(): | |
| task_id = list(TASKS.keys())[0] | |
| task = TASKS[task_id] | |
| score = compute_episode_score(task_id, [task["correct_query"]]) | |
| assert score > 1.0, "Early solve on first attempt should get bonus" | |
| def test_episode_score_no_submissions(): | |
| score = compute_episode_score("easy_missing_from", []) | |
| assert score == 0.0 | |