WIP - wrappers reworking completed, integration test ongoing, (final)…

… next step: settings hints/exposing

diambra/arena/wrappers/observation.py

@@ -328,19 +328,6 @@ def step(self, action):
  obs, reward, terminated, truncated, info = self.env.step(action)
  return self._process_obs(obs), reward, terminated, truncated, info
 
- """
- def rename_key_recursive(dictionary, old_key, new_key):
- if isinstance(dictionary, dict):
- new_dict = {}
- for key, value in dictionary.items():
- if key == old_key:
- key = new_key
- new_dict[key] = rename_key_recursive(value, old_key, new_key)
- return new_dict
- else:
- return dictionary
- """
-
 class FlattenFilterDictObs(gym.ObservationWrapper):
  def __init__(self, env, filter_keys):
  gym.ObservationWrapper.__init__(self, env)

examples/wrappers_options.py

@@ -3,8 +3,7 @@
 
 def main():
  # Environment settings
- #settings = {"n_players": 1, "action_space": SpaceTypes.MULTI_DISCRETE}
- settings = {"n_players": 2, "action_space": (SpaceTypes.MULTI_DISCRETE, SpaceTypes.DISCRETE)}
+ settings = {"n_players": 1, "action_space": SpaceTypes.MULTI_DISCRETE}
 
  # Gym wrappers settings
  wrappers_settings = {}
@@ -61,8 +60,8 @@ def main():
  # optionally exclude images from normalization (False by default)
  # and optionally perform one-hot encoding also on discrete binary variables (False by default)
  wrappers_settings["scale"] = True
- #wrappers_settings["exclude_image_scaling"] = True
- #wrappers_settings["process_discrete_binary"] = True
+ wrappers_settings["exclude_image_scaling"] = True
+ wrappers_settings["process_discrete_binary"] = False
 
  # If to make the observation relative to the agent as a function to its role (P1 or P2) (deactivate by default)
  # i.e.:
@@ -73,14 +72,11 @@ def main():
  # - Under "agent_1", "P1" nesting level becomes "opp" and "P2" becomes "own"
  wrappers_settings["role_relative_observation"] = True
 
- """
  # Flattening observation dictionary and filtering
  # a sub-set of the RAM states
  wrappers_settings["flatten"] = True
- wrappers_settings["filter_keys"] = ["stage", "timer", "agent_0_own_side", "agent_0_opp_side",
- "agent_0_own_health", "agent_0_opp_char",
- "agent_0_action_move", "agent_0_action_attack"]
- """
+ wrappers_settings["filter_keys"] = ["stage", "timer", "action", "own_side", "opp_side",
+ "own_health", "opp_health", "opp_character"]
 
  env = diambra.arena.make("doapp", settings, wrappers_settings, render_mode="human")
 

tests/test_random.py

@@ -86,15 +86,16 @@ def test_random_wrappers_mock(game_id, n_players, action_space, mocker):
  wrappers_settings["clip_rewards"] = False
  wrappers_settings["frame_stack"] = 4
  wrappers_settings["dilation"] = 1
+ wrappers_settings["add_last_action_to_observation"] = True
  wrappers_settings["actions_stack"] = 12
  wrappers_settings["scale"] = True
- wrappers_settings["scale_mod"] = 0
+ wrappers_settings["role_relative_observation"] = True
  wrappers_settings["flatten"] = True
  suffix = ""
  if n_players == 2:
  suffix = "agent_0_"
- wrappers_settings["filter_keys"] = ["stage", "timer", suffix+"own_side", suffix+"opp_side",
- suffix+"opp_char", suffix+"action_move", suffix+"action_attack"]
+ wrappers_settings["filter_keys"] = ["stage", "timer", suffix + "own_side", suffix + "opp_side",
+ suffix + "opp_character", suffix + "action"]
 
  assert func(game_id, n_players, action_space, frame_shape, wrappers_settings,
  no_action_probability, use_mock_env, mocker) == 0
@@ -115,15 +116,16 @@ def test_random_integration(game_id, n_players, action_space, mocker):
  wrappers_settings["clip_rewards"] = False
  wrappers_settings["frame_stack"] = 4
  wrappers_settings["dilation"] = 1
+ wrappers_settings["add_last_action_to_observation"] = True
  wrappers_settings["actions_stack"] = 12
  wrappers_settings["scale"] = True
- wrappers_settings["scale_mod"] = 0
+ wrappers_settings["role_relative_observation"] = True
  wrappers_settings["flatten"] = True
  suffix = ""
  if n_players == 2:
  suffix = "agent_0_"
- wrappers_settings["filter_keys"] = ["stage", "timer", suffix+"own_side", suffix+"opp_side",
- suffix+"opp_char", suffix+"action_move", suffix+"action_attack"]
+ wrappers_settings["filter_keys"] = ["stage", "timer", suffix + "own_side", suffix + "opp_side",
+ suffix + "opp_character", suffix + "action"]
 
  assert func(game_id, n_players, action_space, frame_shape, wrappers_settings,
  no_action_probability, use_mock_env, mocker) == 0

tests/test_recording_settings.py

@@ -66,6 +66,7 @@ def test_settings_recording(game_id ,username, dataset_path, n_players, action_s
  wrappers_settings["frame_shape"] = (128, 128, 1)
  wrappers_settings["reward_normalization"] = True
  wrappers_settings["frame_stack"] = 4
+ wrappers_settings["add_last_action_to_observation"] = True
  wrappers_settings["actions_stack"] = 12
  wrappers_settings["scale"] = True
 

tests/test_speed.py

@@ -76,15 +76,16 @@ def test_speed_wrappers(n_players, mocker):
  wrappers_settings["clip_rewards"] = False
  wrappers_settings["frame_stack"] = 4
  wrappers_settings["dilation"] = 1
+ wrappers_settings["add_last_action_to_observation"] = True
  wrappers_settings["actions_stack"] = 12
  wrappers_settings["scale"] = True
- wrappers_settings["scale_mod"] = 0
+ wrappers_settings["role_relative_observation"] = True
  wrappers_settings["flatten"] = True
 
  suffix = ""
  if n_players == 2:
  suffix = "agent_0_"
- wrappers_settings["filter_keys"] = ["stage", "timer", suffix+"own_side", suffix+"opp_side", suffix+"opp_side",
- suffix+"opp_char", suffix+"action_move", suffix+"action_attack"]
+ wrappers_settings["filter_keys"] = ["stage", "timer", suffix + "own_side", suffix + "opp_side",
+ suffix + "opp_character", suffix + "action"]
 
  assert func(n_players, wrappers_settings, target_speeds[1], mocker) == 0

tests/test_wrappers_settings.py

@@ -26,41 +26,46 @@ def func(settings, wrappers_settings, episode_recording_settings, mocker):
  print("ERROR, ABORTED.")
  return 1
 
-wrappers_settings_var_order = ["no_attack_buttons_combinations", "no_op_max", "sticky_actions", "frame_shape", "reward_normalization",
- "reward_normalization_factor", "clip_rewards", "frame_stack", "dilation",
- "actions_stack", "scale", "flatten", "filter_keys", "wrappers"]
+wrappers_settings_var_order = ["no_op_max", "sticky_actions", "reward_normalization", "reward_normalization_factor",
+ "clip_rewards", "no_attack_buttons_combinations", "frame_shape", "frame_stack", "dilation",
+ "add_last_action_to_observation", "actions_stack", "scale", "role_relative_observation",
+ "flatten", "filter_keys", "wrappers"]
 games_dict = available_games(False)
 
 
 ok_test_parameters = {
- "no_attack_buttons_combinations": [True, False],
  "no_op_max": [0, 2],
  "sticky_actions": [1, 4],
- "frame_shape": [(0, 0, 0), (84, 84, 1), (84, 84, 3), (84, 84, 0)],
  "reward_normalization": [True, False],
  "reward_normalization_factor": [0.2, 0.5],
  "clip_rewards": [True, False],
+ "no_attack_buttons_combinations": [True, False],
+ "frame_shape": [(0, 0, 0), (84, 84, 1), (84, 84, 3), (84, 84, 0)],
  "frame_stack": [1, 5],
  "dilation": [1, 3],
+ "add_last_action_to_observation": [True, False],
  "actions_stack": [1, 6],
  "scale": [True, False],
+ "role_relative_observation": [True, False],
  "flatten": [True, False],
  "filter_keys": [[], ["stage", "own_side"]],
  "wrappers": [[]],
 }
 
 ko_test_parameters = {
- "no_attack_buttons_combinations": [-1],
  "no_op_max": [-1],
  "sticky_actions": [True],
- "frame_shape": [(0, 84, 3), (128, 0, 1)],
  "reward_normalization": ["True"],
  "reward_normalization_factor": [-10],
  "clip_rewards": [0.5],
+ "no_attack_buttons_combinations": [-1],
+ "frame_shape": [(0, 84, 3), (128, 0, 1)],
  "frame_stack": [0],
  "dilation": [0],
+ "add_last_action_to_observation": [10],
  "actions_stack": [-2],
  "scale": [10],
+ "role_relative_observation": [24],
  "flatten": [None],
  "filter_keys": [12],
  "wrappers": ["test"],
@@ -75,9 +80,10 @@ def pytest_generate_tests(metafunc):
 @pytest.mark.parametrize("step_ratio", [1])
 @pytest.mark.parametrize("n_players", [1, 2])
 @pytest.mark.parametrize("action_space", [SpaceTypes.DISCRETE, SpaceTypes.MULTI_DISCRETE])
-def test_wrappers_settings(game_id, step_ratio, n_players, action_space, no_attack_buttons_combinations, no_op_max, sticky_actions,
- frame_shape, reward_normalization, reward_normalization_factor,
- clip_rewards, frame_stack, dilation, actions_stack, scale,
+def test_wrappers_settings(game_id, step_ratio, n_players, action_space, no_op_max, sticky_actions,
+ reward_normalization, reward_normalization_factor, clip_rewards,
+ no_attack_buttons_combinations, frame_shape, frame_stack, dilation,
+ add_last_action_to_observation, actions_stack, scale, role_relative_observation,
  flatten, filter_keys, wrappers, expected, mocker):
 
  # Env settings
@@ -91,17 +97,19 @@ def test_wrappers_settings(game_id, step_ratio, n_players, action_space, no_atta
 
  # Env wrappers settings
  wrappers_settings = {}
- wrappers_settings["no_attack_buttons_combinations"] = no_attack_buttons_combinations
  wrappers_settings["no_op_max"] = no_op_max
  wrappers_settings["sticky_actions"] = sticky_actions
- wrappers_settings["frame_shape"] = frame_shape
  wrappers_settings["reward_normalization"] = reward_normalization
  wrappers_settings["reward_normalization_factor"] = reward_normalization_factor
  wrappers_settings["clip_rewards"] = clip_rewards
+ wrappers_settings["no_attack_buttons_combinations"] = no_attack_buttons_combinations
+ wrappers_settings["frame_shape"] = frame_shape
  wrappers_settings["frame_stack"] = frame_stack
  wrappers_settings["dilation"] = dilation
- wrappers_settings["actions_stack"] = actions_stack
+ wrappers_settings["add_last_action_to_observation"] = add_last_action_to_observation
+ wrappers_settings["actions_stack"] = 1 if add_last_action_to_observation is False and expected == 0 else actions_stack
  wrappers_settings["scale"] = scale
+ wrappers_settings["role_relative_observation"] = role_relative_observation
  wrappers_settings["flatten"] = flatten
  wrappers_settings["filter_keys"] = filter_keys
  wrappers_settings["wrappers"] = wrappers