Fixed update_rewards() to correctly refer to attributes, added test file

Author: bing-j

axelrod/strategies/epsilon_greedy.py

@@ -70,15 +70,20 @@ def update_rewards(self, opponent: Player):
         last_score = game.score(last_round)[0]
 
         # update the expected rewards based on previous play
-        num_plays = (
-            self.cooperations() if last_play == C else self.defections()
-        )
+        if last_play == C:
+            num_plays = self.history.cooperations
+        else:
+            num_plays = self.history.defections
+
         self._rewards[last_play] = self._rewards[last_play] + (
             1 / num_plays
         ) * (last_score - self._rewards[last_play])
 
     def strategy(self, opponent: Player) -> Action:
         """Actual strategy definition that determines player's action."""
+        # if not the first turn
+        if len(self.history) != 0:
+            self.update_rewards(opponent)
 
         # explore
         if self._random.uniform(0.0, 1.0) <= self.epsilon:

axelrod/tests/strategies/test_epsilon_greedy.py

@@ -0,0 +1,28 @@
+"""Tests for the epsilon greedy strategy."""
+
+import axelrod as axl
+
+from .test_player import TestPlayer
+
+C, D = axl.Action.C, axl.Action.D
+
+
+class TestEpsilonGreedy(TestPlayer):
+
+    name = "$\varepsilon$-greedy"
+    player = axl.EpsilonGreedy
+    expected_classifier = {
+        "memory_depth": float("inf"),
+        "stochastic": True,
+        "makes_use_of": {"game"},
+        "long_run_time": False,
+        "inspects_source": False,
+        "manipulates_source": False,
+        "manipulates_state": False,
+    }
+
+    def test_random(self):
+        """Test that strategy is randomly picked (not affected by history)."""
+        opponent = axl.MockPlayer()
+        actions = [(C, C), (C, D), (C, D), (C, C), (C, D)]
+        self.versus_test(opponent, expected_actions=actions, seed=0, init_kwargs={"epsilon": 1})