Upgrade ALE use for latest version of Gymnasium.

PiperOrigin-RevId: 691514007

Author: psc-g

dopamine/discrete_domains/atari_lib.py

@@ -553,7 +553,7 @@ def reset(self):
         environment.
     """
     self.environment.reset()
-    self.lives = self.environment.ale.lives()
+    self.lives = self.environment.env.ale.lives()
     self._fetch_grayscale_observation(self.screen_buffer[0])
     self.screen_buffer[1].fill(0)
     return self._pool_and_resize()
@@ -608,7 +608,7 @@ def step(self, action):
       accumulated_reward += reward
 
       if self.terminal_on_life_loss:
-        new_lives = self.environment.ale.lives()
+        new_lives = self.environment.env.ale.lives()
         is_terminal = game_over or new_lives < self.lives
         self.lives = new_lives
       else:
@@ -639,7 +639,7 @@ def _fetch_grayscale_observation(self, output):
     Returns:
       observation: numpy array, the current observation in grayscale.
     """
-    self.environment.ale.getScreenGrayscale(output)
+    self.environment.env.ale.getScreenGrayscale(output)
     return output
 
   def _pool_and_resize(self):

tests/dopamine/discrete_domains/atari_lib_test.py

@@ -67,32 +67,39 @@ def getScreenGrayscale(self, screen):  # pylint: disable=invalid-name
     screen.fill(self.screen_value)
 
 
+class MockEnvALEWrapper(object):
+  """Mock ALE env wrapper."""
+
+  def __init__(self):
+    self.ale = MockALE()
+
+
 class MockEnvironment(object):
   """Mock environment for testing."""
 
   def __init__(self, screen_size=10, max_steps=10):
     self.max_steps = max_steps
     self.screen_size = screen_size
-    self.ale = MockALE()
+    self.env = MockEnvALEWrapper()
     self.observation_space = np.empty((screen_size, screen_size))
     self.game_over = False
 
   def reset(self):
-    self.ale.screen_value = 10
+    self.env.ale.screen_value = 10
     self.num_steps = 0
     return self.get_observation()
 
   def get_observation(self):
     observation = np.empty((self.screen_size, self.screen_size))
-    return self.ale.getScreenGrayscale(observation)
+    return self.env.ale.getScreenGrayscale(observation)
 
   def step(self, action):
     reward = -1.0 if action > 0 else 1.0
     self.num_steps += 1
     is_terminal = self.num_steps >= self.max_steps
 
     unused = 0
-    self.ale.screen_value = max(0, self.ale.screen_value - 2)
+    self.env.ale.screen_value = max(0, self.env.ale.screen_value - 2)
     return (self.get_observation(), reward, is_terminal, False, unused)
 
   def render(self, mode):