emrekuruu's picture
fix: prevent stalled SpreadsheetBench sessions
81a00cc verified
Raw History Blame Contribute Delete
3.56 kB
"""FastAPI and WebSocket entrypoint for SpreadsheetBench OpenEnv."""
from __future__ import annotations
from collections.abc import Callable
import os
from fastapi import FastAPI
from openenv.core.env_server.http_server import create_app
from openenv.core.env_server.interfaces import Environment
from starlette.types import ASGIApp, Receive, Scope, Send
from starlette.websockets import WebSocketDisconnect
from spreadsheetbench_env.models import (
SpreadsheetBenchAction,
SpreadsheetBenchObservation,
native_tool_specs,
)
from spreadsheetbench_env.server.environment import SpreadsheetBenchEnvironment
_SEND_AFTER_CLOSE = 'Cannot call "send" once a close message has been sent.'
class DisconnectedWebSocketMiddleware:
"""Suppress only OpenEnv errors caused by an already-disconnected client."""
def __init__(self, app: ASGIApp) -> None:
self.app = app
async def __call__(self, scope: Scope, receive: Receive, send: Send) -> None:
try:
await self.app(scope, receive, send)
except WebSocketDisconnect:
if scope["type"] != "websocket":
raise
except RuntimeError as error:
if scope["type"] != "websocket" or str(error) != _SEND_AFTER_CLOSE:
raise
def _configured_capacity() -> int:
raw = os.getenv("MAX_CONCURRENT_ENVS", "1")
try:
capacity = int(raw)
except ValueError as error:
raise ValueError(f"MAX_CONCURRENT_ENVS must be an integer, got {raw!r}") from error
if capacity < 1:
raise ValueError("MAX_CONCURRENT_ENVS must be at least 1")
return capacity
def _configured_default_task_id() -> int | None:
value = os.getenv("DEFAULT_TASK_ID")
if value is None or not value.strip():
return None
try:
return int(value)
except ValueError as error:
raise ValueError(f"DEFAULT_TASK_ID must be an integer, got {value!r}") from error
def _configured_environment_class(
default_task_id: int | None,
) -> type[SpreadsheetBenchEnvironment]:
if default_task_id is None:
return SpreadsheetBenchEnvironment
class ConfiguredSpreadsheetBenchEnvironment(SpreadsheetBenchEnvironment):
def __init__(self) -> None:
super().__init__(default_task_id=default_task_id)
ConfiguredSpreadsheetBenchEnvironment.__name__ = "SpreadsheetBenchEnvironment"
return ConfiguredSpreadsheetBenchEnvironment
def create_spreadsheetbench_app(
max_concurrent_envs: int | None = None,
environment_factory: Callable[[], Environment] | None = None,
) -> FastAPI:
"""Create the pinned OpenEnv app without constructing runtime data eagerly."""
capacity = _configured_capacity() if max_concurrent_envs is None else max_concurrent_envs
if capacity < 1:
raise ValueError("MAX_CONCURRENT_ENVS must be at least 1")
factory = environment_factory or _configured_environment_class(
_configured_default_task_id()
)
app = create_app(
factory,
SpreadsheetBenchAction,
SpreadsheetBenchObservation,
env_name="spreadsheetbench_env",
max_concurrent_envs=capacity,
)
app.add_api_route("/tools", native_tool_specs, methods=["GET"])
app.add_middleware(DisconnectedWebSocketMiddleware)
return app
app = create_spreadsheetbench_app()
def main(host: str = "0.0.0.0", port: int = 8000) -> None:
"""Run the standalone OpenEnv server."""
import uvicorn
uvicorn.run(app, host=host, port=port)
if __name__ == "__main__":
main()