""" 任务槽与认领循环:识别、LibreOffice、Archive 等。 本包约束(硬):LibreOffice 仅 1 槽;HTTP 类 2~3 槽;inbox/outbox 不得与 soffice 串同一阻塞线。 禁止:一把大锁串行所有工单;禁止 wecom:worker:singleton。 """ from __future__ import annotations import logging import threading import time from dataclasses import dataclass, field from typing import Callable logger = logging.getLogger(__name__) @dataclass class SlotPool: """ 进程内有限槽线程池骨架。 用途:限制同类慢活并发(尤其 LibreOffice=1)。 为何不用进程级 Redis 单例:同机只应起一份 Worker 服务;防重入靠任务认领。 """ name: str size: int _sem: threading.BoundedSemaphore = field(init=False, repr=False) def __post_init__(self) -> None: if self.size < 1: raise ValueError(f"槽位数必须 >=1: {self.name}") self._sem = threading.BoundedSemaphore(self.size) def run(self, fn: Callable[[], None], *, label: str = "") -> None: """ 占用一个槽执行 fn;槽满则阻塞等待(调用方应在 Worker 线程,勿在 HTTP 回调线程调用)。 副作用:执行 fn;不写库(骨架)。 """ acquired = self._sem.acquire(timeout=30) if not acquired: logger.warning("槽超时未获取 name=%s label=%s", self.name, label) return try: fn() finally: self._sem.release() @dataclass class WorkerRuntime: """ Worker 运行时:持有分槽与 inbox/outbox 占位线程。 启动时机:worker_app main;停止:进程退出或 stop()。 禁止:把所有任务丢进 LibreOffice 槽。 """ libreoffice: SlotPool http_jobs: SlotPool _stop: threading.Event = field(default_factory=threading.Event) _threads: list[threading.Thread] = field(default_factory=list) @classmethod def from_settings(cls, libreoffice_slots: int, http_job_slots: int) -> "WorkerRuntime": """按配置创建槽;强制 LibreOffice 槽为 1,防止误配双 soffice。""" lo = 1 if libreoffice_slots != 1 else libreoffice_slots if libreoffice_slots != 1: logger.warning( "LibreOffice 槽必须为 1(收到 %s),已强制为 1,避免本机双 soffice 互抢", libreoffice_slots, ) return cls( libreoffice=SlotPool("libreoffice", lo), http_jobs=SlotPool("http_jobs", max(2, http_job_slots)), ) def start(self) -> None: """ 启动 inbox / outbox / 空转巡检线程(骨架:只心跳日志,不消费真实队列)。 副作用:创建守护线程;不连 Redis(骨架)。 """ self._stop.clear() self._threads = [ threading.Thread(target=self._inbox_loop, name="inbox-loop", daemon=True), threading.Thread(target=self._outbox_loop, name="outbox-loop", daemon=True), threading.Thread(target=self._http_slot_idle, name="http-slots-idle", daemon=True), ] for t in self._threads: t.start() logger.info( "WorkerRuntime 已启动 libreoffice_slots=%s http_job_slots=%s", self.libreoffice.size, self.http_jobs.size, ) def stop(self) -> None: """请求停止占位循环;不保证立刻打断正在跑的槽任务(骨架无真实任务)。""" self._stop.set() def _inbox_loop(self) -> None: """inbox 独立线程:后续从 PG/队列认领入站;禁止与 soffice 共用阻塞队列。""" while not self._stop.is_set(): time.sleep(2.0) def _outbox_loop(self) -> None: """outbox 独立线程:后续发送企微;失败重试;过期丢弃不计业务成功。""" while not self._stop.is_set(): time.sleep(2.0) def _http_slot_idle(self) -> None: """证明 HTTP 类槽可并行占用;骨架不跑真实 LLM。""" while not self._stop.is_set(): def _noop() -> None: time.sleep(0.05) self.http_jobs.run(_noop, label="idle-ping") time.sleep(5.0)