mirror of
https://github.com/MODSetter/SurfSense.git
synced 2026-10-22 14:21:35 +00:00
2fb720cd03
- A plugin's server.py may expose a lifespan: the host enters it at startup, warm-ups run in the background, and scraping/lifecycle.py closes the shared sessions, browsers and API clients once at shutdown. - One license cache per process, stale-while-revalidate, instead of one per plugin: the app's 5-minute check no longer pays a Keygen round trip per plugin. Refusals and the 24 h grace work as before. - Kept API clients for Keygen, SearXNG and the captcha solver, which now polls every second. - mcp_server/results.py: tool_result puts each item in the text as one compact JSON line, within 32 KB. opencode reads only a result's text when there is any, so the agent never saw prices, page text or reviews. A request with x-surfsense-text-only: 1 gets no structured copy. - mcp_server/progress.py: stage progress, so a long call never sits silent. - GET and DELETE on /<id>/mcp answer 405: the stateless host has no stream to offer, and each GET held an idle SSE connection open per plugin.
102 lines
3.5 KiB
Python
102 lines
3.5 KiB
Python
"""Progress a tool reports at each stage of a call, so a slow first item does not look dead.
|
|
|
|
A client gives up on a call after a silence (opencode: 200 s), and a progress
|
|
notification is the only message that resets it. Reporting only when an item
|
|
finishes leaves a slow first item silent; reporting each stage, once a second at
|
|
most, keeps the call alive without flooding the stream.
|
|
"""
|
|
|
|
import asyncio
|
|
import logging
|
|
import time
|
|
import weakref
|
|
from dataclasses import dataclass
|
|
|
|
from mcp.server.mcpserver import Context
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
# At most one notification a second per call.
|
|
INTERVAL_SECONDS = 1.0
|
|
|
|
# Read through the module, so a test can move time.
|
|
_clock = time.monotonic
|
|
|
|
|
|
@dataclass
|
|
class _Call:
|
|
"""What a call has reported: whole items done, and stages since the last one."""
|
|
|
|
done: float = 0.0
|
|
stages: int = 0
|
|
total: float | None = None
|
|
sent_at: float | None = None
|
|
# The loop of the call's first stage, the tool's own, which owns its stream.
|
|
loop: asyncio.AbstractEventLoop | None = None
|
|
|
|
def value(self) -> float:
|
|
# Stages between two items count as fractions that never reach the next
|
|
# item, so the value rises with every notification, as MCP requires.
|
|
return self.done + self.stages / (self.stages + 1)
|
|
|
|
|
|
# By the context's id, removed when the context is: a Context cannot be hashed.
|
|
_calls: dict[int, _Call] = {}
|
|
|
|
|
|
async def stage(
|
|
ctx: Context | None,
|
|
message: str,
|
|
*,
|
|
done: int | None = None,
|
|
total: int | None = None,
|
|
) -> None:
|
|
"""Report `message`, with `done` of `total` items when given, as a progress notification.
|
|
|
|
Sent at most once a second per call, except the stage whose `done` first reaches
|
|
`total`, which is always sent. Never raises: progress is not worth failing a
|
|
call for. A no-op without a context or when the client asked for no progress.
|
|
A stage reported from another loop, such as the browser loop, is sent on the
|
|
loop of the call's first stage.
|
|
"""
|
|
if ctx is None:
|
|
return
|
|
try:
|
|
call = _calls.get(id(ctx))
|
|
if call is None:
|
|
call = _calls[id(ctx)] = _Call()
|
|
# Before its id can be reused.
|
|
weakref.finalize(ctx, _calls.pop, id(ctx), None)
|
|
if total is not None:
|
|
call.total = total
|
|
# Only the stage that first reaches the total: a plugin may count past a
|
|
# per-URL total, and the items after it are limited like any other.
|
|
last = (
|
|
done is not None
|
|
and call.total is not None
|
|
and call.done < call.total <= done
|
|
)
|
|
if done is not None and done > call.value():
|
|
call.done, call.stages = done, 0
|
|
else:
|
|
last = False
|
|
call.stages += 1
|
|
now = _clock()
|
|
if (
|
|
not last
|
|
and call.sent_at is not None
|
|
and now - call.sent_at < INTERVAL_SECONDS
|
|
):
|
|
return
|
|
call.sent_at = now
|
|
loop = asyncio.get_running_loop()
|
|
call.loop = call.loop or loop
|
|
if call.loop is loop:
|
|
await ctx.report_progress(call.value(), call.total, message)
|
|
elif not call.loop.is_closed():
|
|
# Not awaited: the browser loop's work need not wait on the stream.
|
|
asyncio.run_coroutine_threadsafe(
|
|
ctx.report_progress(call.value(), call.total, message), call.loop
|
|
)
|
|
except Exception as error:
|
|
logger.debug("[progress] not reported: %r", error)
|