Update environment.py

environment.py

@@ -506,21 +506,23 @@ class FocusFlowEnvironment:
             feedback_parts.append(f"Unknown action '{atype}': -0.05.")
 
         return reward, " | ".join(feedback_parts)
-   '''For each uncompleted deadline, it calculates how close you are to missing it. At 50+ steps away → pressure = 0.0. At 0 steps away → pressure=1.0.
-  Returns the highest pressure across all deadlines. 
-  This number appears in the observation so the LLM knows when to stop chatting and start studying.'''
-    def _compute_deadline_pressure(self) -> float:
+        
+     def _compute_deadline_pressure(self) -> float:
+        # For each uncompleted deadline, it calculates how close you are to missing it.
+        # At 50+ steps away → pressure = 0.0. At 0 steps away → pressure=1.0.
+        # Returns the highest pressure across all deadlines. 
+        # This number appears in the observation so the LLM knows when to stop chatting and start studying.
         if not self.day_context.pending_deadlines:
             return 0.0
         pressures = []
         for dl in self.day_context.pending_deadlines:
-            if dl["completed"]:
+            if dl.get("completed", False):
                 continue
             steps_left = dl["due_step"] - self.step_count
             if steps_left <= 0:
                 pressures.append(1.0)
             else:
-                pressures.append(max(0.0, 1.0 - steps_left / 50.0))
+                pressures.append(max(0.0, 1.0 - (steps_left / 50.0)))
         return max(pressures) if pressures else 0.0
 
     # ── Public OpenEnv API ────────────────────────────────────────────────────