Update environment.py
Browse files- environment.py +7 -7
environment.py
CHANGED
|
@@ -242,14 +242,14 @@ class CodeReviewEnv:
|
|
| 242 |
|
| 243 |
# ===================================================================
|
| 244 |
def _get_observation(self) -> EnhancedObservation:
|
| 245 |
-
|
| 246 |
-
|
| 247 |
-
|
| 248 |
-
|
| 249 |
-
|
| 250 |
author_response = ""
|
| 251 |
|
| 252 |
-
|
| 253 |
code_snippet=self._current_code,
|
| 254 |
last_tool_output=self._test_results or "",
|
| 255 |
author_response=author_response, # ← fixed
|
|
@@ -279,7 +279,7 @@ class CodeReviewEnv:
|
|
| 279 |
|
| 280 |
bug_description=self._bug_description,
|
| 281 |
comments_count=len(self._comments),
|
| 282 |
-
|
| 283 |
# ===================================================================
|
| 284 |
def _compute_dense_reward(
|
| 285 |
self,
|
|
|
|
| 242 |
|
| 243 |
# ===================================================================
|
| 244 |
def _get_observation(self) -> EnhancedObservation:
|
| 245 |
+
"""Return COMPLETE Markov state."""
|
| 246 |
+
# Compute author response: only after comment/question/fix does the author actually speak
|
| 247 |
+
if self._last_action_type in ("write_comment", "ask_question", "propose_fix"):
|
| 248 |
+
author_response = self._test_results or ""
|
| 249 |
+
else:
|
| 250 |
author_response = ""
|
| 251 |
|
| 252 |
+
return EnhancedObservation(
|
| 253 |
code_snippet=self._current_code,
|
| 254 |
last_tool_output=self._test_results or "",
|
| 255 |
author_response=author_response, # ← fixed
|
|
|
|
| 279 |
|
| 280 |
bug_description=self._bug_description,
|
| 281 |
comments_count=len(self._comments),
|
| 282 |
+
)
|
| 283 |
# ===================================================================
|
| 284 |
def _compute_dense_reward(
|
| 285 |
self,
|