refactor(utils): default reward target_field to 'rewards' and drop 'reward_config' alias (#2474)

scripts/visualize_task_env.py

@@ -24,12 +24,11 @@
 from __future__ import annotations
 
 import argparse
-import dataclasses
 import sys
 import time
 from collections.abc import Sequence
 from pathlib import Path
-from typing import TYPE_CHECKING, Any, cast, get_args, get_origin, get_type_hints
+from typing import TYPE_CHECKING, Any, cast
 
 import numpy as np
 from unisim.backend.base import CameraCfg
@@ -51,45 +50,6 @@
     from unilab.base.scene import SceneCfg
 
 
-def _build_reward_stub(env_cfg_cls: type) -> dict[str, Any] | None:
-    """Generate a minimal reward_config dict for the given env cfg class."""
-    try:
-        type_hints = get_type_hints(env_cfg_cls)
-    except Exception:
-        return None
-    reward_type = type_hints.get("reward_config")
-    if reward_type is None:
-        return None
-    if get_origin(reward_type) is not None:
-        non_none = [a for a in get_args(reward_type) if a is not type(None)]
-        if not non_none:
-            return None
-        reward_type = non_none[0]
-    if not dataclasses.is_dataclass(reward_type):
-        return None
-
-    stub: dict[str, Any] = {}
-    for f in dataclasses.fields(reward_type):
-        if f.default is not dataclasses.MISSING:
-            continue
-        if f.default_factory is not dataclasses.MISSING:  # type: ignore[misc]
-            continue
-        type_str = str(f.type)
-        if "dict" in type_str:
-            stub[f.name] = {}
-        elif "list" in type_str or "tuple" in type_str or "Sequence" in type_str:
-            stub[f.name] = []
-        elif "bool" in type_str:
-            stub[f.name] = False
-        elif "float" in type_str or "int" in type_str:
-            stub[f.name] = 0.0
-        elif "str" in type_str:
-            stub[f.name] = ""
-        else:
-            stub[f.name] = None
-    return stub
-
-
 def _parse_args(argv: Sequence[str] | None = None) -> argparse.Namespace:
     parser = argparse.ArgumentParser(
         description="Visualize an RL task's training environment with zero actions."
@@ -291,12 +251,7 @@ def _build_env_cfg_override(task_name: str) -> dict[str, Any]:
             f"Task '{task_name}' is not registered. "
             f"Available: {sorted(registry.list_registered_envs())}"
         )
-    env_cfg = registry.materialize_env_config(task_name)
-    override: dict[str, Any] = {}
-    reward_stub = _build_reward_stub(type(env_cfg))
-    if reward_stub is not None:
-        override["reward_config"] = reward_stub
-    return override
+    return {}
 
 
 def _print_backend_scene(env) -> None:

src/unilab/utils/reward.py

@@ -19,9 +19,7 @@ def _to_reward_dict(value: object, *, error_message: str) -> RewardDict:
     return cast(RewardDict, resolved)
 
 
-def resolve_reward_dict(
-    cfg: DictConfig, *, target_field: str = "reward_config"
-) -> dict[str, RewardDict]:
+def resolve_reward_dict(cfg: DictConfig, *, target_field: str = "rewards") -> dict[str, RewardDict]:
     """Extract and validate the reward config from a composed Hydra config.
 
     Args:
@@ -33,12 +31,10 @@ def resolve_reward_dict(
 
     Raises:
         ValueError: If the reward config is missing or empty, or the
-            ``target_field`` is not a supported override key.
+            ``target_field`` is not the supported ``"rewards"`` override key.
     """
-    if target_field not in {"reward_config", "rewards"}:
-        raise ValueError(
-            f"Reward config target_field must be 'reward_config' or 'rewards', got {target_field!r}"
-        )
+    if target_field != "rewards":
+        raise ValueError(f"Reward config target_field must be 'rewards', got {target_field!r}")
     reward_cfg = OmegaConf.select(cfg, "reward")
     if not reward_cfg:
         raise ValueError(

tests/config/test_reward_injection.py

@@ -24,13 +24,29 @@ def test_resolve_reward_dict_reads_task_reward():
             overrides=["task=go2_joystick_flat/mujoco"],
         )
 
-    reward_dict = resolve_reward_dict(cfg)["reward_config"]
+    reward_dict = resolve_reward_dict(cfg)["rewards"]
 
     assert reward_dict["tracking_lin_vel"]["weight"] == 1.0
     assert reward_dict["tracking_ang_vel"]["weight"] == 0.2
     assert reward_dict["tracking_lin_vel"]["func"].endswith("track_lin_vel")
 
 
+def test_resolve_reward_dict_rejects_reward_config_target():
+    """The removed 'reward_config' override key must fail closed."""
+    import pytest
+
+    from unilab.utils.reward import resolve_reward_dict
+
+    with initialize(config_path="../../src/unilab/conf/ppo", version_base="1.3"):
+        cfg = compose(
+            config_name="config",
+            overrides=["task=go2_joystick_flat/mujoco"],
+        )
+
+    with pytest.raises(ValueError, match="target_field must be 'rewards'"):
+        resolve_reward_dict(cfg, target_field="reward_config")
+
+
 def test_reward_config_conversion():
     """Test reward config materializes into manager reward terms via registry."""
     from unilab.base import registry