File size: 12,462 Bytes
12a1ac5 5c7b7df 12a1ac5 8aedea9 12a1ac5 8aedea9 12a1ac5 03465ea 12a1ac5 03465ea 12a1ac5 03465ea 8aedea9 03465ea 12a1ac5 03465ea 12a1ac5 03465ea 12a1ac5 03465ea 12a1ac5 03465ea 12a1ac5 03465ea 12a1ac5 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 | """
server/env.py
CrossSessionContinuityEnv — main MCPEnvironment subclass.
Implements the Gym-style API:
env.reset() → starts episode, returns Session 1 observation
env.step() → action → (obs, reward, done, info)
env.state() → current env state dict
Two-session architecture:
Session 1: agent reads code, writes code, runs tests, writes handoff note.
Session 2: agent calls parse_handoff(), reads, writes, runs tests, submits.
Key enforcement:
- parse_handoff() MUST be called before any file access in Session 2.
- HandoffValidator rejects malformed notes (not penalises — retry possible).
- Filesystem wiped on session transition.
- Dynamic step limits by difficulty.
- Retry budget for invalid actions.
"""
from __future__ import annotations
from dataclasses import dataclass, field
from typing import Any, Dict, Optional
from uuid import uuid4
from server.task_generator import TaskGenerator
from server.session_manager import SessionManager
from server.sandbox import Sandbox
from server.rewards.rubric import ContinuityRubric
from server.rewards.auxiliary import AuxiliaryRewarder
from server.handoff_validator import HandoffValidator
try:
from models import ContinuityAction, ContinuityObservation
except ImportError:
try:
from ..models import ContinuityAction, ContinuityObservation
except ImportError:
from models import ContinuityAction, ContinuityObservation
# ---------------------------------------------------------------------------
# OpenEnv base class — openenv-core package
# ---------------------------------------------------------------------------
try:
from openenv.core.env_server.interfaces import Environment as _EnvBase
from openenv.core.env_server.types import State
_HAS_OPENENV = True
except ImportError:
class State: # type: ignore[no-redef]
def __init__(self, **kwargs):
for k, v in kwargs.items(): setattr(self, k, v)
class _EnvBase: # type: ignore[no-redef]
pass
_HAS_OPENENV = False
# ---------------------------------------------------------------------------
# Constants
# ---------------------------------------------------------------------------
STEP_LIMITS: Dict[str, int] = {
"easy": 20,
"medium": 35,
"hard": 55,
}
# ---------------------------------------------------------------------------
# Action dataclass
# ---------------------------------------------------------------------------
@dataclass
class Action:
tool: str
path: str = ""
content: str = ""
args: Dict[str, Any] = field(default_factory=dict)
# ---------------------------------------------------------------------------
# Environment
# ---------------------------------------------------------------------------
class CrossSessionContinuityEnv(_EnvBase):
"""
RL environment for cross-session coding continuity.
Inherits from openenv.core.env_server.interfaces.Environment.
Implements OpenEnv Gym-style: reset / step / state (property) / close.
Registered tools: read_file, write_file, run_tests,
write_handoff, parse_handoff, submit.
"""
def __init__(self, difficulty: str = "medium"):
assert difficulty in STEP_LIMITS, f"Invalid difficulty: {difficulty}"
self.difficulty = difficulty
self.task_gen = TaskGenerator(difficulty)
self.session_mgr = SessionManager()
self.sandbox = Sandbox(timeout=10)
self.rubric = ContinuityRubric()
self.aux = AuxiliaryRewarder()
self.validator = HandoffValidator()
self.step_limit = STEP_LIMITS[difficulty]
# Episode state (populated in reset)
self.task = None
self.session = 1
self.handoff = None
self.step_count = 0
self.invalid_action_count = 0
self.retry_budget = 3
self.s1_test_history = []
self.s2_edit_history = []
self.handoff_parsed = False
self.s2_failed_runs = 0
# ------------------------------------------------------------------
# Gym-style API
# ------------------------------------------------------------------
def reset(
self,
task_id: Optional[str] = None,
seed: Optional[int] = None,
) -> Dict[str, Any]:
"""Start a new episode. Returns Session 1 observation."""
self.task = self.task_gen.sample(task_id, seed=seed)
self.session = 1
self.handoff = None
self.step_count = 0
self.invalid_action_count = 0
self.retry_budget = 3
self.s1_test_history = []
self.s2_edit_history = []
self.handoff_parsed = False
self.s2_failed_runs = 0
return {
"session": 1,
"task": self.task.description,
"starter_code": self.task.starter_code,
"message": "Session 1 started. Complete what you can, then call write_handoff().",
"step_limit": self.step_limit,
}
def step(self, action: Action) -> Dict[str, Any]:
"""
Execute one agent action. Returns observation dict with optional
'done', 'reward', and 'auxiliary_reward' keys.
"""
self.step_count += 1
# Late-step warning in Session 1
if self.session == 1 and self.step_count > self.step_limit:
return {
"warning": "Step limit reached. Call write_handoff() now or episode terminates.",
"penalty": -0.1,
"session": 1,
"done": False,
}
# Invalid action guard
if not self._is_valid_action(action):
self.invalid_action_count += 1
self.retry_budget -= 1
if self.retry_budget <= 0:
return {"done": True, "reward": 0.0, "error": "Retry budget exhausted."}
return {
"error": f"Invalid action '{action.tool}' in session {self.session}.",
"retries_left": self.retry_budget,
"done": False,
}
# Dispatch by tool
return self._dispatch(action)
@property
def state(self) -> State:
"""OpenEnv required: return current State object."""
return State(
session=self.session,
step_count=self.step_count,
step_limit=self.step_limit,
handoff_written=self.handoff is not None,
handoff_length=len(self.handoff.split()) if self.handoff else 0,
difficulty=self.difficulty,
invalid_actions=self.invalid_action_count,
task_id=self.task.task_id if self.task else None,
)
# ------------------------------------------------------------------
# Tool dispatch
# ------------------------------------------------------------------
def _dispatch(self, action: Action) -> Dict[str, Any]:
t = action.tool
if t == "read_file":
return self._handle_read_file(action)
if t == "write_file":
return self._handle_write_file(action)
if t == "run_tests":
return self._handle_run_tests(action)
if t == "write_handoff":
return self._handle_write_handoff(action)
if t == "parse_handoff":
return self._handle_parse_handoff(action)
if t == "submit":
return self._handle_submit(action)
return {"error": f"Unknown tool: {t}", "done": False}
def _handle_read_file(self, action: Action) -> Dict[str, Any]:
if self.session == 2 and not self.handoff_parsed:
return {"error": "Call parse_handoff() before accessing files in Session 2.", "done": False}
content = self.task.files.get(action.path, f"File not found: {action.path}")
return {"output": content, "session": self.session, "done": False}
def _handle_write_file(self, action: Action) -> Dict[str, Any]:
prev = self.task.files.get(action.path, "")
self.task.files[action.path] = action.content
if self.session == 2:
self.s2_edit_history.append({
"path": action.path,
"prev": prev,
"new": action.content,
})
return {"output": f"Written to {action.path}", "session": self.session, "done": False}
def _handle_run_tests(self, action: Action) -> Dict[str, Any]: # noqa: ARG002
result = self.sandbox.run_tests(self.task.files, self.task.test_code)
if self.session == 1:
self.s1_test_history.append(result.passed)
aux = self.aux.s1_reward(result, self.task)
return {
"output": result.summary,
"passed": result.passed,
"total": result.total,
"auxiliary_reward": aux,
"session": 1,
"done": False,
}
else:
if result.passed == 0:
self.s2_failed_runs += 1
return {
"output": result.summary,
"passed": result.passed,
"total": result.total,
"session": 2,
"done": False,
}
def _handle_write_handoff(self, action: Action) -> Dict[str, Any]:
if self.session != 1:
return {"error": "write_handoff only available in Session 1.", "done": False}
validation = self.validator.validate(action.content)
if not validation.valid:
return {
"error": f"Handoff rejected: {validation.reason}",
"required": self.validator.REQUIRED_SECTIONS,
"done": False,
}
self.handoff = action.content
self.session = 2
self.handoff_parsed = False
self.task = self.session_mgr.transition(self.task)
self.retry_budget = 3 # fresh budget for Session 2
return {
"session": 2,
"message": "Session 2 started. Call parse_handoff() first.",
"done": False,
}
def _handle_parse_handoff(self, action: Action) -> Dict[str, Any]: # noqa: ARG002
if self.session != 2:
return {"error": "parse_handoff only available in Session 2.", "done": False}
self.handoff_parsed = True
return {"output": self.handoff, "session": 2, "done": False}
def _handle_submit(self, action: Action) -> Dict[str, Any]: # noqa: ARG002
if self.session != 2:
return {"error": "submit only available in Session 2.", "done": False}
visible = self.sandbox.run_tests(self.task.files, self.task.test_code)
hidden = self.sandbox.run_tests(self.task.files, self.task.hidden_test_code)
reward_breakdown = self.rubric.score(
visible_results=visible,
hidden_results=hidden,
handoff=self.handoff,
s2_edit_history=self.s2_edit_history,
s2_failed_runs=self.s2_failed_runs,
invalid_actions=self.invalid_action_count,
)
return {
"done": True,
"reward": reward_breakdown.total,
"breakdown": {
"test_score": reward_breakdown.test_score,
"quality_score": reward_breakdown.quality_score,
"linearity_score": reward_breakdown.linearity_score,
"rewrite_penalty": reward_breakdown.rewrite_penalty,
"action_penalty": reward_breakdown.action_penalty,
},
"visible_summary": visible.summary,
"hidden_summary": hidden.summary,
}
def close(self) -> None:
"""OpenEnv required: teardown. No-op for this environment."""
pass
# ------------------------------------------------------------------
# Helpers
# ------------------------------------------------------------------
def _is_valid_action(self, action: Action) -> bool:
s1_tools = {"read_file", "write_file", "run_tests", "write_handoff"}
s2_tools = {"parse_handoff", "read_file", "write_file", "run_tests", "submit"}
return action.tool in (s1_tools if self.session == 1 else s2_tools)
|