55aafbae9a
- Modified loadChatHistory() to check for active project before fetching all sessions - When active project exists, use project.sessions instead of fetching from API - Added detailed console logging to debug session filtering - This prevents ALL sessions from appearing in every project's sidebar Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
384 lines
13 KiB
Python
384 lines
13 KiB
Python
import asyncio
|
|
import logging
|
|
import signal
|
|
import threading
|
|
from dataclasses import dataclass, field
|
|
from datetime import datetime, timezone
|
|
from typing import (
|
|
Any,
|
|
AsyncIterable,
|
|
Awaitable,
|
|
Callable,
|
|
Coroutine,
|
|
Iterator,
|
|
Mapping,
|
|
Optional,
|
|
Set,
|
|
Union,
|
|
)
|
|
|
|
import anyio
|
|
from starlette.background import BackgroundTask
|
|
from starlette.concurrency import iterate_in_threadpool
|
|
from starlette.datastructures import MutableHeaders
|
|
from starlette.responses import Response
|
|
from starlette.types import Receive, Scope, Send, Message
|
|
|
|
from sse_starlette.event import ServerSentEvent, ensure_bytes
|
|
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
|
|
@dataclass
|
|
class _ShutdownState:
|
|
"""Per-thread state for shutdown coordination.
|
|
|
|
Issue #152 fix: Uses threading.local() instead of ContextVar to ensure
|
|
one watcher per thread rather than one per async context.
|
|
"""
|
|
|
|
events: Set[anyio.Event] = field(default_factory=set)
|
|
watcher_started: bool = False
|
|
|
|
|
|
# Each thread gets its own shutdown state (one event loop per thread typically)
|
|
_thread_state = threading.local()
|
|
|
|
|
|
def _get_shutdown_state() -> _ShutdownState:
|
|
"""Get or create shutdown state for the current thread."""
|
|
state = getattr(_thread_state, "shutdown_state", None)
|
|
if state is None:
|
|
state = _ShutdownState()
|
|
_thread_state.shutdown_state = state
|
|
return state
|
|
|
|
|
|
def _get_uvicorn_server():
|
|
"""
|
|
Try to get uvicorn Server instance via signal handler introspection.
|
|
|
|
When uvicorn registers signal handlers, they're bound methods on the Server instance.
|
|
We can retrieve the Server from the handler's __self__ attribute.
|
|
|
|
Returns None if:
|
|
- Not running under uvicorn
|
|
- Signal handler isn't a bound method
|
|
- Any introspection fails
|
|
"""
|
|
try:
|
|
handler = signal.getsignal(signal.SIGTERM)
|
|
if hasattr(handler, "__self__"):
|
|
server = handler.__self__
|
|
if hasattr(server, "should_exit"):
|
|
return server
|
|
except Exception:
|
|
pass
|
|
return None
|
|
|
|
|
|
async def _shutdown_watcher() -> None:
|
|
"""
|
|
Poll for shutdown and broadcast to all events in this context.
|
|
|
|
One watcher runs per thread (event loop). Checks two shutdown sources:
|
|
1. AppStatus.should_exit - set when our monkey-patch works
|
|
2. uvicorn Server.should_exit - via signal handler introspection (Issue #132 fix)
|
|
|
|
When either becomes True, signals all registered events.
|
|
"""
|
|
state = _get_shutdown_state()
|
|
uvicorn_server = _get_uvicorn_server()
|
|
|
|
try:
|
|
while True:
|
|
# Check our flag (monkey-patch worked or manually set)
|
|
if AppStatus.should_exit:
|
|
break
|
|
# Check uvicorn's flag directly (monkey-patch failed - Issue #132)
|
|
if (
|
|
AppStatus.enable_automatic_graceful_drain
|
|
and uvicorn_server is not None
|
|
and uvicorn_server.should_exit
|
|
):
|
|
AppStatus.should_exit = True # Sync state for consistency
|
|
break
|
|
await anyio.sleep(0.5)
|
|
|
|
# Shutdown detected - broadcast to all waiting events
|
|
for event in list(state.events):
|
|
event.set()
|
|
finally:
|
|
# Allow watcher to be restarted if loop is reused
|
|
state.watcher_started = False
|
|
|
|
|
|
def _ensure_watcher_started_on_this_loop() -> None:
|
|
"""Ensure the shutdown watcher is running for this thread (event loop)."""
|
|
state = _get_shutdown_state()
|
|
if not state.watcher_started:
|
|
state.watcher_started = True
|
|
try:
|
|
loop = asyncio.get_running_loop()
|
|
loop.create_task(_shutdown_watcher())
|
|
except RuntimeError:
|
|
# No running loop - shouldn't happen in normal use
|
|
state.watcher_started = False
|
|
|
|
|
|
class SendTimeoutError(TimeoutError):
|
|
pass
|
|
|
|
|
|
class AppStatus:
|
|
"""Helper to capture a shutdown signal from Uvicorn so we can gracefully terminate SSE streams."""
|
|
|
|
should_exit = False
|
|
enable_automatic_graceful_drain = True
|
|
original_handler: Optional[Callable] = None
|
|
|
|
@staticmethod
|
|
def disable_automatic_graceful_drain():
|
|
"""
|
|
Prevent automatic SSE stream termination on server shutdown.
|
|
|
|
WARNING: When disabled, you MUST set AppStatus.should_exit = True
|
|
at some point during shutdown, or streams will never close and the
|
|
server will hang indefinitely (or until uvicorn's graceful shutdown
|
|
timeout expires).
|
|
"""
|
|
AppStatus.enable_automatic_graceful_drain = False
|
|
|
|
@staticmethod
|
|
def enable_automatic_graceful_drain_mode():
|
|
"""
|
|
Re-enable automatic SSE stream termination on server shutdown.
|
|
|
|
This restores the default behavior where SIGTERM triggers immediate
|
|
stream draining. Call this to undo a previous call to
|
|
disable_automatic_graceful_drain().
|
|
"""
|
|
AppStatus.enable_automatic_graceful_drain = True
|
|
|
|
@staticmethod
|
|
def handle_exit(*args, **kwargs):
|
|
if AppStatus.enable_automatic_graceful_drain:
|
|
AppStatus.should_exit = True
|
|
if AppStatus.original_handler is not None:
|
|
AppStatus.original_handler(*args, **kwargs)
|
|
|
|
|
|
try:
|
|
from uvicorn.main import Server
|
|
|
|
AppStatus.original_handler = Server.handle_exit
|
|
Server.handle_exit = AppStatus.handle_exit # type: ignore
|
|
except ImportError:
|
|
logger.debug(
|
|
"Uvicorn not installed. Graceful shutdown on server termination disabled."
|
|
)
|
|
|
|
Content = Union[str, bytes, dict, ServerSentEvent, Any]
|
|
SyncContentStream = Iterator[Content]
|
|
AsyncContentStream = AsyncIterable[Content]
|
|
ContentStream = Union[AsyncContentStream, SyncContentStream]
|
|
|
|
|
|
class EventSourceResponse(Response):
|
|
"""
|
|
Streaming response that sends data conforming to the SSE (Server-Sent Events) specification.
|
|
"""
|
|
|
|
DEFAULT_PING_INTERVAL = 15
|
|
DEFAULT_SEPARATOR = "\r\n"
|
|
|
|
def __init__(
|
|
self,
|
|
content: ContentStream,
|
|
status_code: int = 200,
|
|
headers: Optional[Mapping[str, str]] = None,
|
|
media_type: str = "text/event-stream",
|
|
background: Optional[BackgroundTask] = None,
|
|
ping: Optional[int] = None,
|
|
sep: Optional[str] = None,
|
|
ping_message_factory: Optional[Callable[[], ServerSentEvent]] = None,
|
|
data_sender_callable: Optional[
|
|
Callable[[], Coroutine[None, None, None]]
|
|
] = None,
|
|
send_timeout: Optional[float] = None,
|
|
client_close_handler_callable: Optional[
|
|
Callable[[Message], Awaitable[None]]
|
|
] = None,
|
|
) -> None:
|
|
# Validate separator
|
|
if sep not in (None, "\r\n", "\r", "\n"):
|
|
raise ValueError(f"sep must be one of: \\r\\n, \\r, \\n, got: {sep}")
|
|
self.sep = sep or self.DEFAULT_SEPARATOR
|
|
|
|
# If content is sync, wrap it for async iteration
|
|
if isinstance(content, AsyncIterable):
|
|
self.body_iterator = content
|
|
else:
|
|
self.body_iterator = iterate_in_threadpool(content)
|
|
|
|
self.status_code = status_code
|
|
self.media_type = self.media_type if media_type is None else media_type
|
|
self.background = background
|
|
self.data_sender_callable = data_sender_callable
|
|
self.send_timeout = send_timeout
|
|
|
|
# Build SSE-specific headers.
|
|
_headers = MutableHeaders()
|
|
if headers is not None: # pragma: no cover
|
|
_headers.update(headers)
|
|
|
|
# "The no-store response directive indicates that any caches of any kind (private or shared)
|
|
# should not store this response."
|
|
# -- https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Cache-Control
|
|
# allow cache control header to be set by user to support fan out proxies
|
|
# https://www.fastly.com/blog/server-sent-events-fastly
|
|
|
|
_headers.setdefault("Cache-Control", "no-store")
|
|
# mandatory for servers-sent events headers
|
|
_headers["Connection"] = "keep-alive"
|
|
_headers["X-Accel-Buffering"] = "no"
|
|
self.init_headers(_headers)
|
|
|
|
self.ping_interval = self.DEFAULT_PING_INTERVAL if ping is None else ping
|
|
self.ping_message_factory = ping_message_factory
|
|
|
|
self.client_close_handler_callable = client_close_handler_callable
|
|
|
|
self.active = True
|
|
# https://github.com/sysid/sse-starlette/pull/55#issuecomment-1732374113
|
|
self._send_lock = anyio.Lock()
|
|
|
|
@property
|
|
def ping_interval(self) -> Union[int, float]:
|
|
return self._ping_interval
|
|
|
|
@ping_interval.setter
|
|
def ping_interval(self, value: Union[int, float]) -> None:
|
|
if not isinstance(value, (int, float)):
|
|
raise TypeError("ping interval must be int")
|
|
if value < 0:
|
|
raise ValueError("ping interval must be greater than 0")
|
|
self._ping_interval = value
|
|
|
|
def enable_compression(self, force: bool = False) -> None:
|
|
raise NotImplementedError("Compression is not supported for SSE streams.")
|
|
|
|
async def _stream_response(self, send: Send) -> None:
|
|
"""Send out SSE data to the client as it becomes available in the iterator."""
|
|
await send(
|
|
{
|
|
"type": "http.response.start",
|
|
"status": self.status_code,
|
|
"headers": self.raw_headers,
|
|
}
|
|
)
|
|
|
|
async for data in self.body_iterator:
|
|
chunk = ensure_bytes(data, self.sep)
|
|
logger.debug("chunk: %s", chunk)
|
|
with anyio.move_on_after(self.send_timeout) as cancel_scope:
|
|
await send(
|
|
{"type": "http.response.body", "body": chunk, "more_body": True}
|
|
)
|
|
|
|
if cancel_scope and cancel_scope.cancel_called:
|
|
if hasattr(self.body_iterator, "aclose"):
|
|
await self.body_iterator.aclose()
|
|
raise SendTimeoutError()
|
|
|
|
async with self._send_lock:
|
|
self.active = False
|
|
await send({"type": "http.response.body", "body": b"", "more_body": False})
|
|
|
|
async def _listen_for_disconnect(self, receive: Receive) -> None:
|
|
"""Watch for a disconnect message from the client."""
|
|
while self.active:
|
|
message = await receive()
|
|
if message["type"] == "http.disconnect":
|
|
self.active = False
|
|
logger.debug("Got event: http.disconnect. Stop streaming.")
|
|
if self.client_close_handler_callable:
|
|
await self.client_close_handler_callable(message)
|
|
break
|
|
|
|
@staticmethod
|
|
async def _listen_for_exit_signal() -> None:
|
|
"""Wait for shutdown signal via the shared watcher."""
|
|
if AppStatus.should_exit:
|
|
return
|
|
|
|
_ensure_watcher_started_on_this_loop()
|
|
|
|
state = _get_shutdown_state()
|
|
event = anyio.Event()
|
|
state.events.add(event)
|
|
|
|
try:
|
|
# Double-check after registration
|
|
if AppStatus.should_exit:
|
|
return
|
|
await event.wait()
|
|
finally:
|
|
state.events.discard(event)
|
|
|
|
async def _ping(self, send: Send) -> None:
|
|
"""Periodically send ping messages to keep the connection alive on proxies.
|
|
- frequenccy ca every 15 seconds.
|
|
- Alternatively one can send periodically a comment line (one starting with a ':' character)
|
|
"""
|
|
while self.active:
|
|
await anyio.sleep(self._ping_interval)
|
|
sse_ping = (
|
|
self.ping_message_factory()
|
|
if self.ping_message_factory
|
|
else ServerSentEvent(
|
|
comment=f"ping - {datetime.now(timezone.utc)}", sep=self.sep
|
|
)
|
|
)
|
|
ping_bytes = ensure_bytes(sse_ping, self.sep)
|
|
logger.debug("ping: %s", ping_bytes)
|
|
|
|
async with self._send_lock:
|
|
if self.active:
|
|
await send(
|
|
{
|
|
"type": "http.response.body",
|
|
"body": ping_bytes,
|
|
"more_body": True,
|
|
}
|
|
)
|
|
|
|
async def __call__(self, scope: Scope, receive: Receive, send: Send) -> None:
|
|
"""Entrypoint for Starlette's ASGI contract. We spin up tasks:
|
|
- _stream_response to push events
|
|
- _ping to keep the connection alive
|
|
- _listen_for_exit_signal to respond to server shutdown
|
|
- _listen_for_disconnect to respond to client disconnect
|
|
"""
|
|
async with anyio.create_task_group() as task_group:
|
|
# https://trio.readthedocs.io/en/latest/reference-core.html#custom-supervisors
|
|
async def cancel_on_finish(coro: Callable[[], Awaitable[None]]):
|
|
await coro()
|
|
task_group.cancel_scope.cancel()
|
|
|
|
task_group.start_soon(cancel_on_finish, lambda: self._stream_response(send))
|
|
task_group.start_soon(cancel_on_finish, lambda: self._ping(send))
|
|
task_group.start_soon(cancel_on_finish, self._listen_for_exit_signal)
|
|
|
|
if self.data_sender_callable:
|
|
task_group.start_soon(self.data_sender_callable)
|
|
|
|
# Wait for the client to disconnect last
|
|
task_group.start_soon(
|
|
cancel_on_finish, lambda: self._listen_for_disconnect(receive)
|
|
)
|
|
|
|
if self.background is not None:
|
|
await self.background()
|