强化学习第一版

tests/test_planner_revision.py

@@ -5,12 +5,13 @@ import math
 import pytest
 
 from mcm26b.config import ProblemConfig
+from mcm26b.harness import LocalSimTransport, LocalSimulator
 from mcm26b.planner import ClearCertificate, DirectionalExtension
 from mcm26b.planner.bayes_loop import BayesLoop, Task
 from mcm26b.planner.seven import validate_seven_coverage
 from mcm26b.protocol.client import SimulatorClient
 from mcm26b.protocol.rules import DIRECTION, NO_SIGNAL
-from mcm26b.sim import LocalSimulator, LocalSimTransport, Source
+from referee import Source
 
 
 def setup(sources=(), cls=BayesLoop, config=ProblemConfig()):

tests/test_rl.py

@@ -0,0 +1,102 @@
+"""RL 扩展验收：特征/动作布局、训练环境闭环与模型接口守卫。
+
+RL 依赖属于可选组 [dependency-groups] rl；未安装 numpy/gymnasium 时
+全部跳过（运行闭环不依赖它们）。
+"""
+
+import math
+import unittest
+from pathlib import Path
+
+pytest = None
+try:
+    import pytest
+    numpy = pytest.importorskip("numpy")
+    pytest.importorskip("gymnasium")
+except Exception:  # pragma: no cover - 无 RL 依赖环境
+    numpy = None
+
+if numpy is not None:
+    from mcm26b.belief import CandidateRegion
+    from mcm26b.config import ProblemConfig
+    from mcm26b.rl._constants import GLOBAL_DIM, NUM_ACTIONS, OBS_DIM, PER_ACTION_DIM
+    from mcm26b.rl.env import LocalizationEnv
+    from mcm26b.rl.features import build_observation, candidate_actions
+    from mcm26b.rl.policy import default_model_path
+
+CONFIG = ProblemConfig()
+STATION = (0.0, 0.0)
+BEARING = 30.0
+
+
+def _region():
+    region = CandidateRegion.initial(STATION, CONFIG)
+    region.clip_wedge(STATION, BEARING, 1.005)
+    return region
+
+
+@unittest.skipIf(numpy is None, "RL 依赖组未安装")
+class FeatureLayoutTests(unittest.TestCase):
+    def test_action_candidates_have_stable_count_and_order(self):
+        region = _region()
+        first = candidate_actions(region, STATION, CONFIG, [])
+        self.assertEqual(len(first), NUM_ACTIONS)
+        second = candidate_actions(region, STATION, CONFIG, [])
+        self.assertEqual(first, second)
+        self.assertEqual(first[0], region.mec()[0])   # 0 号候选 = MEC 圆心
+
+    def test_observation_dimension_matches_model_contract(self):
+        region = _region()
+        obs = build_observation(region, STATION, 3, CONFIG, problem=3,
+                                blocked=[], virtual_time_s=0.0)
+        self.assertEqual(obs.shape, (OBS_DIM,))
+        self.assertEqual(OBS_DIM, GLOBAL_DIM + NUM_ACTIONS * PER_ACTION_DIM)
+
+    def test_blocked_changes_observation_but_not_layout(self):
+        region = _region()
+        clean = build_observation(region, STATION, 3, CONFIG, problem=4,
+                                  blocked=[], virtual_time_s=0.0)
+        blocked = build_observation(region, STATION, 3, CONFIG, problem=4,
+                                    blocked=[(100.0, 0.0)], virtual_time_s=0.0)
+        self.assertEqual(clean.shape, blocked.shape)
+        self.assertFalse(numpy.allclose(clean, blocked))
+
+
+@unittest.skipIf(numpy is None, "RL 依赖组未安装")
+class TrainingEnvironmentTests(unittest.TestCase):
+    def test_q3_episode_reset_and_step_shapes(self):
+        env = LocalizationEnv(problem=3)
+        obs, info = env.reset(seed=1)
+        self.assertEqual(obs.shape, (OBS_DIM,))
+        self.assertIn("source", info)
+        obs, reward, terminated, truncated, info = env.step(0)
+        self.assertEqual(obs.shape, (OBS_DIM,))
+        self.assertIsInstance(reward, float)
+        self.assertFalse(terminated and truncated)
+
+    def test_q4_directional_episode_reports_blocked_or_fallback(self):
+        env = LocalizationEnv(problem=4, directional_probability=1.0)
+        obs, info = env.reset(seed=7)
+        self.assertTrue(info["source"]["directional"])
+        terminated = truncated = False
+        for _ in range(env.max_episode_steps + 2):
+            obs, reward, terminated, truncated, info = env.step(
+                env.action_space.sample())
+            if terminated or truncated:
+                break
+        self.assertTrue(terminated or truncated)
+        # 定向分支：要么清除，要么走过后备清除/截断语义。
+        self.assertIn("fallback_attempts", info)
+
+    def test_missing_model_raises_clear_error(self):
+        from mcm26b.rl.policy import PointSelectionPolicy
+        with self.assertRaises(RuntimeError):
+            PointSelectionPolicy("output/rl/does-not-exist.zip")
+
+    def test_default_model_paths(self):
+        self.assertEqual(default_model_path(3), Path("output/rl/q3_ppo.zip"))
+        self.assertEqual(default_model_path(4), Path("output/rl/q4_ppo.zip"))
+
+
+if __name__ == "__main__":
+    unittest.main()