854 lines
37 KiB
Python
854 lines
37 KiB
Python
"""通用任务管理框架：任务类型注册 + 设备分组 + 任务计划 + 定时调度 + 重试 + 持久化。
|
||
|
||
设计目标：可扩展，未来加非抖音任务只需注册新的 Task 类。
|
||
|
||
核心概念：
|
||
TaskType — 任务类型（如"抖音养号"），可注册，含默认参数和 worker 工厂
|
||
DeviceGroup — 设备分组，持久化到 SQLite（core.models.DeviceGroup）
|
||
TaskJob — 任务计划（类型+目标+参数+调度+重试），持久化到 SQLite（core.models.TaskJob）
|
||
TaskManager — 统管调度器、分组、任务、运行实例、状态
|
||
|
||
定时：APScheduler BackgroundScheduler，cron 触发后为每个目标设备起 worker。
|
||
支持三种调度模式：
|
||
once — 手动/立即执行，不注册 cron
|
||
cron — 仅注册启动 cron（到点启动 worker）
|
||
cron_stop — 注册启动 cron + 停止 cron（到点启动，到点停止）
|
||
重试：单设备 worker 失败后按 retry.max_attempts 重试，间隔 retry.delay。
|
||
并发：同一 serial 同时只允许一个 worker，避免冲突。
|
||
"""
|
||
import os
|
||
import json
|
||
import time
|
||
import uuid
|
||
import threading
|
||
from datetime import datetime
|
||
from concurrent.futures import ThreadPoolExecutor, as_completed
|
||
|
||
from apscheduler.schedulers.background import BackgroundScheduler
|
||
from apscheduler.triggers.cron import CronTrigger
|
||
|
||
from config import DATA_DIR
|
||
from core.logger import get_logger
|
||
from core.models import db, DeviceGroup as GroupRow, TaskJob as JobRow
|
||
from core import device_pool
|
||
from .stf_client import STFClient, DeviceOfflineError
|
||
from .adb_helper import get_foreground_app, adb_connect_light, adb_disconnect
|
||
from .device_worker import (
|
||
get_all_worker_status, _update_status, _remove_worker,
|
||
_WORKERS, _WORKERS_LOCK,
|
||
start_watchdog, stop_watchdog,
|
||
)
|
||
from tasks import list_task_types, get_task_class
|
||
|
||
_log = get_logger("core.tm")
|
||
|
||
# 错峰启动间隔（秒）：批量触发时设备逐个开始连接，避免 adb 全局锁串行堆积。
|
||
# 100 台 × 0.2s = 20s 全部开始启动。
|
||
_START_STAGGER_SEC = 0.2
|
||
|
||
|
||
def _in_run_window(schedule, now=None):
|
||
"""是否在当前运行窗口内。
|
||
|
||
schedule.window = {"start": "HH:MM", "end": "HH:MM"}，每天重复；
|
||
窗口外（定时触发 + 手动执行）任务不会启动。未配置/非法配置视为不限制。
|
||
支持跨午夜（如 21:00-09:00 = 晚上 9 点运行到次日早 9 点）。
|
||
"""
|
||
win = (schedule or {}).get("window") or {}
|
||
start, end = win.get("start", ""), win.get("end", "")
|
||
if not start or not end:
|
||
return True
|
||
try:
|
||
cur = (now or datetime.now()).hour * 60 + (now or datetime.now()).minute
|
||
s = int(start.split(":")[0]) * 60 + int(start.split(":")[1])
|
||
e = int(end.split(":")[0]) * 60 + int(end.split(":")[1])
|
||
except (ValueError, AttributeError, IndexError):
|
||
return True # 配置非法按不限制处理
|
||
if s == e:
|
||
return True # 起止相同视为不限制
|
||
if s < e:
|
||
return s <= cur < e
|
||
return cur >= s or cur < e # 跨午夜
|
||
|
||
|
||
def _next_run_time(schedule, now=None):
|
||
"""任务下次真正执行的时间（考虑运行窗口）。
|
||
|
||
用 CronTrigger.get_next_fire_time 从当前时间向后找触发点，
|
||
跳过运行窗口外的触发点（最多找 200 次防死循环）。
|
||
返回 datetime 或 None（无 cron / 配置非法）。
|
||
"""
|
||
sched = schedule or {}
|
||
if sched.get("mode") not in ("cron", "cron_stop"):
|
||
return None
|
||
cron = sched.get("cron", "")
|
||
if not cron:
|
||
return None
|
||
try:
|
||
trigger = CronTrigger.from_crontab(cron)
|
||
except Exception:
|
||
return None
|
||
now = now or datetime.now()
|
||
win = sched.get("window") or {}
|
||
if not win.get("start") or not win.get("end"):
|
||
return trigger.get_next_fire_time(None, now)
|
||
fire = trigger.get_next_fire_time(None, now)
|
||
for _ in range(200):
|
||
if fire is None:
|
||
return None
|
||
if _in_run_window({"window": win}, fire):
|
||
return fire
|
||
fire = trigger.get_next_fire_time(fire, fire)
|
||
return None
|
||
|
||
|
||
# ================== 设备分组 ==================
|
||
class DeviceGroup:
|
||
def __init__(self, name, serials=None, description=""):
|
||
self.name = name
|
||
self.serials = serials or []
|
||
self.description = description
|
||
|
||
def to_dict(self):
|
||
return {"name": self.name, "serials": self.serials, "description": self.description}
|
||
|
||
@classmethod
|
||
def from_dict(cls, d):
|
||
return cls(d["name"], d.get("serials", []), d.get("description", ""))
|
||
|
||
|
||
# ================== 任务计划 ==================
|
||
class TaskJob:
|
||
"""一个任务计划：什么任务、跑哪些设备、什么参数、何时跑、失败重试策略。"""
|
||
|
||
def __init__(self, job_id=None, name="", task_type="douyin_nurture",
|
||
target=None, params=None, schedule=None, retry=None, enabled=True):
|
||
self.id = job_id or uuid.uuid4().hex[:8]
|
||
self.name = name
|
||
self.task_type = task_type
|
||
# target: {"mode": "all"|"group"|"serial", "group_name": "", "serial": ""}
|
||
self.target = target or {"mode": "all"}
|
||
self.params = params or {}
|
||
self.schedule = schedule or {"mode": "once"} # {"mode":"once"|"cron", "cron":"0 9 * * *"}
|
||
self.retry = retry or {"max_attempts": 1, "delay": 60}
|
||
self.enabled = enabled
|
||
|
||
def to_dict(self):
|
||
return {"id": self.id, "name": self.name, "task_type": self.task_type,
|
||
"target": self.target, "params": self.params,
|
||
"schedule": self.schedule, "retry": self.retry, "enabled": self.enabled}
|
||
|
||
@classmethod
|
||
def from_dict(cls, d):
|
||
return cls(d["id"], d["name"], d["task_type"], d.get("target"),
|
||
d.get("params"), d.get("schedule"), d.get("retry"), d.get("enabled", True))
|
||
|
||
def resolve_serials(self, manager):
|
||
"""根据 target 解析出实际要跑的 serial 列表。
|
||
|
||
数据源：core.device_pool（本地清单 + adb 在线状态），不再查 STF。
|
||
"""
|
||
mode = self.target.get("mode", "all")
|
||
if mode == "serial":
|
||
serials = [self.target["serial"]]
|
||
elif mode == "group":
|
||
g = manager.groups.get(self.target.get("group_name"))
|
||
serials = list(g.serials) if g else []
|
||
else:
|
||
# all：默认返回池内在线设备（单实例互斥由 _run_with_retry 的 _running 保证）；
|
||
# 抢占模式返回全部在线设备（含运行中，执行时抢占）
|
||
if self.params.get("preempt"):
|
||
try:
|
||
return device_pool.list_online()
|
||
except Exception:
|
||
return []
|
||
try:
|
||
return device_pool.list_ready()
|
||
except Exception:
|
||
return []
|
||
# 离线自动跳过（serial/group 模式）：跳过本机 adb 不可达的设备，
|
||
# 避免反复尝试连接后报"设备离线"。默认开启，可在任务编辑器取消勾选。
|
||
if serials and self.params.get("skip_offline", True):
|
||
try:
|
||
online = set(device_pool.list_online())
|
||
except Exception:
|
||
online = None # 查询失败时不过滤，维持原行为
|
||
if online is not None:
|
||
kept, skipped = [], []
|
||
for s in serials:
|
||
if s in online:
|
||
kept.append(s)
|
||
else:
|
||
skipped.append((s, "离线(未连接)"))
|
||
if skipped:
|
||
detail = ", ".join(f"{s}({why})" for s, why in skipped)
|
||
_log.info(f"任务 {self.name} 跳过 {len(skipped)} 台设备: {detail}")
|
||
serials = kept
|
||
return serials
|
||
|
||
|
||
# ================== 前台 App 扫描器 ==================
|
||
class _ForegroundScanner:
|
||
"""前台 App 扫描器：手动触发，获取所有在线设备的前台 App。
|
||
|
||
设计原则：**不打扰设备**，扫描过程不会让设备退出当前 App。
|
||
|
||
策略（按设备状态区分获取逻辑）：
|
||
- worker 运行中设备：用已有 remote_adb_url 直接查询（已有 adb 连接，无额外开销）
|
||
- 空闲设备：直接 adb connect <serial> → 查询 → adb disconnect（绕过 STF，不打扰设备）
|
||
- 被别人占用的设备：标记 "他人占用"
|
||
- STF occupy/release 会打扰设备（可能退回桌面），绝不使用
|
||
|
||
serial 格式为 IP:5555（设备本身的 adb 网络地址），可直接 adb connect。
|
||
adb connect/disconnect 只建立/断开调试连接，不影响设备 UI。
|
||
dumpsys window 只读取窗口状态，不执行任何操作。
|
||
"""
|
||
|
||
def __init__(self):
|
||
self._cache = {} # serial -> app_name
|
||
self._cache_lock = threading.Lock()
|
||
self._scanning = threading.Event() # 标记是否正在扫描
|
||
self._last_scan_time = 0 # 上次扫描完成时间戳
|
||
self._last_scan_count = 0 # 上次扫描设备数
|
||
|
||
def stop(self):
|
||
"""兼容 TaskManager.shutdown() 调用。"""
|
||
pass
|
||
|
||
def get(self, serial):
|
||
"""读取缓存的前台 App。"""
|
||
with self._cache_lock:
|
||
return self._cache.get(serial)
|
||
|
||
@property
|
||
def is_scanning(self):
|
||
return self._scanning.is_set()
|
||
|
||
@property
|
||
def last_scan_time(self):
|
||
return self._last_scan_time
|
||
|
||
def scan_once(self):
|
||
"""手动触发一次扫描（后台线程执行，不阻塞调用者）。
|
||
|
||
如果已有扫描在进行中，直接返回 False。
|
||
"""
|
||
if self._scanning.is_set():
|
||
return False
|
||
t = threading.Thread(target=self._scan_all, name="fg-scan-once", daemon=True)
|
||
t.start()
|
||
return True
|
||
|
||
def _scan_all(self):
|
||
"""扫描所有在线设备（后台线程执行）。
|
||
|
||
按设备归属分两类处理：
|
||
1. worker 运行中：用已有 remote_adb_url 直接查询（无额外开销）
|
||
2. 空闲设备：返回"空闲"——不主动连接。IP:5555 的 adb transport
|
||
与 STF provider 共享，外部 connect/disconnect 会让 STF 误判
|
||
设备离线并触发重连（迁移期仍保留此约束，摘除 STF 后可放开）
|
||
"""
|
||
self._scanning.set()
|
||
_log.info("前台 App 扫描已启动")
|
||
try:
|
||
online = set(device_pool.list_online())
|
||
except Exception as e:
|
||
_log.error("前台 App 扫描: 获取在线设备失败: %s", e)
|
||
self._scanning.clear()
|
||
return
|
||
|
||
worker_status = {w["serial"]: w for w in get_all_worker_status()}
|
||
|
||
# 分类设备
|
||
have_conn = {} # serial -> remote_adb_url（worker 运行中，已有 adb 连接）
|
||
free_serials = [] # 空闲设备
|
||
|
||
for serial in online:
|
||
w = worker_status.get(serial, {})
|
||
url = w.get("remote_adb_url")
|
||
if url and w.get("status") in ("running", "connecting"):
|
||
# 1. 设备正在执行任务，已有 adb 连接，直接查询
|
||
have_conn[serial] = url
|
||
else:
|
||
# 2. 空闲设备
|
||
free_serials.append(serial)
|
||
|
||
results = {}
|
||
_log.info("前台 App 扫描分类: 运行中=%d, 空闲=%d",
|
||
len(have_conn), len(free_serials))
|
||
|
||
# 1. worker 运行中设备：用已有 remote_adb_url 查询（并发 10）
|
||
if have_conn:
|
||
with ThreadPoolExecutor(max_workers=min(10, len(have_conn))) as pool:
|
||
futures = {pool.submit(get_foreground_app, url): s
|
||
for s, url in have_conn.items()}
|
||
for fut in as_completed(futures, timeout=15):
|
||
s = futures[fut]
|
||
try:
|
||
results[s] = fut.result()
|
||
except Exception:
|
||
results[s] = None
|
||
|
||
# 2. 空闲设备：不打扰，直接返回"空闲"
|
||
for s in free_serials:
|
||
results[s] = "空闲"
|
||
|
||
# 更新缓存
|
||
with self._cache_lock:
|
||
for s, app in results.items():
|
||
self._cache[s] = app
|
||
|
||
self._last_scan_time = time.time()
|
||
self._last_scan_count = len(results)
|
||
self._scanning.clear()
|
||
_log.info("前台 App 扫描完成: %d 台设备", len(results))
|
||
|
||
|
||
# ================== 任务管理器 ==================
|
||
class TaskManager:
|
||
def __init__(self, stf_client=None, app=None):
|
||
self.stf = stf_client or STFClient()
|
||
self.app = app # Flask app，用于 db context
|
||
self.scheduler = BackgroundScheduler(timezone="Asia/Shanghai")
|
||
self.scheduler.start()
|
||
start_watchdog() # 启动心跳看门狗
|
||
self.groups = {} # name -> DeviceGroup(内存业务对象)
|
||
self.jobs = {} # id -> TaskJob(内存业务对象)
|
||
self._running = {} # serial -> {"worker", "job_id", "started_at", "attempt"}
|
||
self._stop_requested = set() # serial 集合：用户请求停止，阻止后续重试
|
||
self._lock = threading.Lock()
|
||
self._fg_scanner = _ForegroundScanner()
|
||
self._load()
|
||
|
||
def _db(self):
|
||
"""获取 db context。task_manager 在后台线程跑，需自己推 app context。"""
|
||
if self.app is None:
|
||
raise RuntimeError("TaskManager 未关联 Flask app，无法访问数据库")
|
||
return self.app.app_context()
|
||
|
||
# ---- 持久化（SQLite，替代旧 JSON 文件）----
|
||
def _load(self):
|
||
"""从数据库加载分组和任务到内存。"""
|
||
try:
|
||
with self._db():
|
||
for row in GroupRow.query.all():
|
||
g = DeviceGroup(row.name, row.get_serials(), row.description or "")
|
||
self.groups[g.name] = g
|
||
dirty = False # 标记是否有任务参数需要写回
|
||
for row in JobRow.query.all():
|
||
j = TaskJob(job_id=row.id, name=row.name, task_type=row.task_type,
|
||
target=row.get_target(), params=row.get_params(),
|
||
schedule=row.get_schedule(), retry=row.get_retry(),
|
||
enabled=row.enabled)
|
||
# 数据规整：移除已废弃的 action 配置（如抖音的 comment 已删除）
|
||
if j.task_type == "douyin_nurture":
|
||
actions = j.params.get("actions", {})
|
||
if "comment" in actions:
|
||
del actions["comment"]
|
||
row.set_params(j.params)
|
||
dirty = True
|
||
_log.info(f"任务 {j.name}({j.id}): 已移除废弃的 comment 配置")
|
||
self.jobs[j.id] = j
|
||
if dirty:
|
||
db.session.commit()
|
||
_log.info("已将规整后的任务参数写回数据库")
|
||
_log.info(f"从数据库加载 {len(self.groups)} 个分组, {len(self.jobs)} 个任务")
|
||
except Exception as e:
|
||
_log.error(f"从数据库加载失败: {e}")
|
||
# 恢复启用的 cron 任务
|
||
for job in self.jobs.values():
|
||
if job.enabled and job.schedule.get("mode") in ("cron", "cron_stop"):
|
||
self._add_cron(job)
|
||
|
||
def _save_groups(self):
|
||
"""保存所有分组到数据库（upsert）。"""
|
||
with self._db():
|
||
for g in self.groups.values():
|
||
row = GroupRow.query.filter_by(name=g.name).first()
|
||
if not row:
|
||
row = GroupRow(name=g.name)
|
||
db.session.add(row)
|
||
row.set_serials(g.serials)
|
||
row.description = g.description
|
||
db.session.commit()
|
||
|
||
def _save_jobs(self):
|
||
"""保存所有任务到数据库（upsert）。"""
|
||
with self._db():
|
||
for j in self.jobs.values():
|
||
row = JobRow.query.get(j.id)
|
||
if not row:
|
||
row = JobRow(id=j.id)
|
||
db.session.add(row)
|
||
row.name = j.name
|
||
row.task_type = j.task_type
|
||
row.set_target(j.target)
|
||
row.set_params(j.params)
|
||
row.set_schedule(j.schedule)
|
||
row.set_retry(j.retry)
|
||
row.enabled = j.enabled
|
||
db.session.commit()
|
||
|
||
# ---- 分组管理 ----
|
||
def add_group(self, name, serials, description=""):
|
||
g = DeviceGroup(name, serials, description)
|
||
self.groups[name] = g
|
||
self._save_groups()
|
||
return g
|
||
|
||
def update_group(self, name, serials=None, description=None):
|
||
g = self.groups.get(name)
|
||
if not g:
|
||
return None
|
||
if serials is not None:
|
||
g.serials = serials
|
||
if description is not None:
|
||
g.description = description
|
||
self._save_groups()
|
||
return g
|
||
|
||
def delete_group(self, name):
|
||
if name in self.groups:
|
||
del self.groups[name]
|
||
self._save_groups()
|
||
# 同上：显式删数据库行，避免删除的分组重启后复活
|
||
with self._db():
|
||
row = GroupRow.query.filter_by(name=name).first()
|
||
if row:
|
||
db.session.delete(row)
|
||
db.session.commit()
|
||
return True
|
||
return False
|
||
|
||
# ---- 任务计划管理 ----
|
||
def add_job(self, name, task_type, target, params, schedule, retry, enabled=True):
|
||
job = TaskJob(name=name, task_type=task_type, target=target, params=params,
|
||
schedule=schedule, retry=retry, enabled=enabled)
|
||
self.jobs[job.id] = job
|
||
if enabled and schedule.get("mode") in ("cron", "cron_stop"):
|
||
self._add_cron(job)
|
||
self._save_jobs()
|
||
return job
|
||
|
||
def update_job(self, job_id, **fields):
|
||
job = self.jobs.get(job_id)
|
||
if not job:
|
||
return None
|
||
# 更新前先移除旧 cron
|
||
was_cron = job.enabled and job.schedule.get("mode") in ("cron", "cron_stop")
|
||
if was_cron:
|
||
self._remove_cron(job_id)
|
||
for k, v in fields.items():
|
||
if v is not None and hasattr(job, k):
|
||
setattr(job, k, v)
|
||
if job.enabled and job.schedule.get("mode") in ("cron", "cron_stop"):
|
||
self._add_cron(job)
|
||
self._save_jobs()
|
||
return job
|
||
|
||
def delete_job(self, job_id):
|
||
self._remove_cron(job_id)
|
||
if job_id in self.jobs:
|
||
del self.jobs[job_id]
|
||
self._save_jobs()
|
||
# _save_jobs 只做 upsert 不会删行：必须显式删数据库行，
|
||
# 否则删除的任务重启后会从数据库重新加载回来
|
||
with self._db():
|
||
row = JobRow.query.get(job_id)
|
||
if row:
|
||
db.session.delete(row)
|
||
db.session.commit()
|
||
return True
|
||
return False
|
||
|
||
def toggle_job(self, job_id, enabled):
|
||
job = self.jobs.get(job_id)
|
||
if not job:
|
||
return None
|
||
job.enabled = enabled
|
||
if enabled and job.schedule.get("mode") in ("cron", "cron_stop"):
|
||
self._add_cron(job)
|
||
else:
|
||
self._remove_cron(job_id)
|
||
self._save_jobs()
|
||
return job
|
||
|
||
# ---- APScheduler 集成 ----
|
||
def _cron_id(self, job_id, kind="start"):
|
||
"""生成 cron job id。kind: start=启动 cron，stop=停止 cron。"""
|
||
return f"job_{job_id}_{kind}"
|
||
|
||
def _add_cron(self, job):
|
||
"""注册定时任务。支持三种模式：
|
||
- once: 不注册 cron
|
||
- cron: 只注册启动 cron（到点启动 worker）
|
||
- cron_stop: 注册启动 cron + 停止 cron（到点启动，到点停止）
|
||
"""
|
||
mode = job.schedule.get("mode", "once")
|
||
if mode not in ("cron", "cron_stop"):
|
||
return
|
||
|
||
start_cron = job.schedule.get("cron", "")
|
||
if start_cron:
|
||
try:
|
||
trigger = CronTrigger.from_crontab(start_cron)
|
||
except Exception as e:
|
||
_log.error(f"启动 cron 解析失败 {start_cron}: {e}")
|
||
else:
|
||
self.scheduler.add_job(self._on_cron_trigger, trigger, args=[job.id],
|
||
id=self._cron_id(job.id, "start"),
|
||
replace_existing=True)
|
||
_log.info(f"已注册启动定时 {job.name}({job.id}) cron={start_cron}")
|
||
|
||
if mode == "cron_stop":
|
||
stop_cron = job.schedule.get("stop_cron", "")
|
||
if stop_cron:
|
||
try:
|
||
trigger = CronTrigger.from_crontab(stop_cron)
|
||
except Exception as e:
|
||
_log.error(f"停止 cron 解析失败 {stop_cron}: {e}")
|
||
else:
|
||
self.scheduler.add_job(self._on_cron_stop, trigger, args=[job.id],
|
||
id=self._cron_id(job.id, "stop"),
|
||
replace_existing=True)
|
||
_log.info(f"已注册停止定时 {job.name}({job.id}) stop_cron={stop_cron}")
|
||
|
||
def _remove_cron(self, job_id):
|
||
"""移除该任务的所有 cron job（启动 + 停止）。"""
|
||
for kind in ("start", "stop"):
|
||
try:
|
||
self.scheduler.remove_job(self._cron_id(job_id, kind))
|
||
except Exception:
|
||
pass
|
||
|
||
def _on_cron_trigger(self, job_id):
|
||
"""cron 触发：启动该任务的所有目标设备 worker。"""
|
||
job = self.jobs.get(job_id)
|
||
if not job:
|
||
return
|
||
if not _in_run_window(job.schedule):
|
||
win = job.schedule.get("window") or {}
|
||
_log.info(f"定时触发 {job.name}({job.id}) 跳过：当前不在运行窗口内 "
|
||
f"（{win.get('start','')}-{win.get('end','')}）")
|
||
return
|
||
_log.info(f"定时触发: {job.name}({job.id})")
|
||
self._run_job(job)
|
||
|
||
def _on_cron_stop(self, job_id):
|
||
"""停止 cron 触发：停止该任务正在运行的所有 worker。
|
||
|
||
只停本 job 启动的 worker，不影响其他任务。
|
||
"""
|
||
job = self.jobs.get(job_id)
|
||
if not job:
|
||
return
|
||
with self._lock:
|
||
items = [(s, info) for s, info in self._running.items()
|
||
if info.get("job_id") == job_id]
|
||
stopped = []
|
||
for serial, info in items:
|
||
w = info.get("worker")
|
||
if w and w.is_alive():
|
||
w.stop()
|
||
stopped.append(serial)
|
||
_log.info(f"定时停止 {job.name}({job.id}): 停止 {len(stopped)} 台设备 {stopped}")
|
||
|
||
def next_run_of(self, job):
|
||
"""任务下次真正执行的时间（考虑运行窗口）。停用/无 cron 返回 None。"""
|
||
if not job or not job.enabled:
|
||
return None
|
||
return _next_run_time(job.schedule)
|
||
|
||
def run_job_now(self, job_id):
|
||
"""立即执行任务（手动触发）。运行窗口外拒绝启动。"""
|
||
job = self.jobs.get(job_id)
|
||
if not job:
|
||
return {"ok": False, "error": "任务不存在"}
|
||
if not _in_run_window(job.schedule):
|
||
win = job.schedule.get("window") or {}
|
||
return {"ok": False,
|
||
"error": f"当前不在运行窗口内（{win.get('start', '')}-{win.get('end', '')}），任务未启动"}
|
||
# 在独立线程跑，不阻塞调用方
|
||
t = threading.Thread(target=self._run_job, args=(job,), daemon=True)
|
||
t.start()
|
||
return {"ok": True, "msg": f"任务 {job.name} 已触发"}
|
||
|
||
def _run_job(self, job):
|
||
"""执行任务：为每个目标设备起 worker（含重试循环）。
|
||
|
||
大量设备同时启动会触发 adb connect 全局锁串行 + STF occupy 并发风暴，
|
||
这里按 _START_STAGGER_SEC 间隔逐个启动，分摊连接压力（100 台 × 0.2s = 20s）。
|
||
"""
|
||
serials = job.resolve_serials(self)
|
||
if not serials:
|
||
_log.warning(f"任务 {job.name} 无可用设备")
|
||
return
|
||
task_cls = get_task_class(job.task_type)
|
||
if not task_cls:
|
||
_log.error(f"未知任务类型: {job.task_type}")
|
||
return
|
||
task = task_cls()
|
||
max_attempts = max(1, job.retry.get("max_attempts", 1))
|
||
delay = job.retry.get("delay", 60)
|
||
|
||
for idx, serial in enumerate(serials):
|
||
# 每台设备一个重试循环线程，互不影响；错峰延迟在各自线程内等待
|
||
t = threading.Thread(target=self._run_with_retry,
|
||
args=(task, serial, job, max_attempts, delay,
|
||
idx * _START_STAGGER_SEC), daemon=True)
|
||
t.start()
|
||
|
||
def _run_with_retry(self, task, serial, job, max_attempts, delay, start_delay=0):
|
||
"""单设备任务执行 + 重试。
|
||
|
||
异常分类：
|
||
DeviceOfflineError — 设备掉线，立即放弃不重试（换设备也没用）
|
||
其他异常 — 按 max_attempts 重试
|
||
用户停止（stop_device）会加入 _stop_requested，阻止任何后续重试。
|
||
"""
|
||
# 错峰启动：在各自线程内等待，分摊批量启动的连接/占用压力
|
||
if start_delay > 0:
|
||
time.sleep(start_delay)
|
||
try:
|
||
for attempt in range(1, max_attempts + 1):
|
||
# 用户已请求停止 → 不再启动新 attempt
|
||
with self._lock:
|
||
if serial in self._stop_requested:
|
||
_log.info(f"{serial} 用户已请求停止，取消重试 (job={job.name})")
|
||
return
|
||
# 同一 serial 同时只能一个 worker；不同任务可配置抢占
|
||
preempt = False
|
||
preempted_job = None # 被抢占的原任务 id（抢占结束后归还）
|
||
with self._lock:
|
||
if serial in self._running:
|
||
cur = self._running[serial]
|
||
if cur.get("job_id") == job.id:
|
||
# 同一任务重复触发：跳过（原行为）
|
||
_log.warning(f"{serial} 已有任务在跑，跳过 (job={job.name})")
|
||
return
|
||
if not job.params.get("preempt"):
|
||
# 不同任务且本任务未开启抢占：跳过
|
||
_log.warning(f"{serial} 已有任务在跑，跳过 (job={job.name})")
|
||
return
|
||
preempt = True
|
||
preempted_job = cur.get("job_id")
|
||
if preempt:
|
||
# 抢占：锁外停止该设备上的其他任务（stop_device 内部拿同一把锁，
|
||
# 在锁内调用会死锁！），等其释放后接管
|
||
_log.warning(f"{serial} 抢占：停止任务 {preempted_job} 后执行 {job.name}")
|
||
self.stop_device(serial)
|
||
deadline = time.time() + 30
|
||
while time.time() < deadline:
|
||
with self._lock:
|
||
if serial not in self._running:
|
||
break
|
||
time.sleep(0.5)
|
||
else:
|
||
_log.warning(f"{serial} 抢占超时（旧任务 30s 未退出），跳过 (job={job.name})")
|
||
return
|
||
with self._lock:
|
||
self._running[serial] = {"job_id": job.id, "started_at": time.time(),
|
||
"attempt": attempt, "task_type": job.task_type,
|
||
"preempted_job": preempted_job}
|
||
_update_status(serial, task_job=job.name, attempt=attempt,
|
||
max_attempts=max_attempts)
|
||
|
||
worker = None
|
||
try:
|
||
worker = task.create_worker(self.stf, serial, job.params)
|
||
with self._lock:
|
||
self._running[serial]["worker"] = worker
|
||
_log.info(f"{serial} 开始任务 {job.name} (第{attempt}/{max_attempts}次)")
|
||
worker.start()
|
||
worker.join() # 等待 worker 结束
|
||
# worker 正常结束（done 或被 stop）
|
||
with self._lock:
|
||
self._running.pop(serial, None)
|
||
# 判断是否成功：看 status
|
||
with _WORKERS_LOCK:
|
||
st = _WORKERS.get(serial, {}).get("status")
|
||
# 清除 task_job 标记，避免前端误判"运行中"
|
||
_update_status(serial, task_job="")
|
||
if st == "done":
|
||
_log.info(f"{serial} 任务 {job.name} 成功完成")
|
||
return
|
||
# 用户请求停止（无论 attempt 第几次、status 是什么）→ 不重试
|
||
with self._lock:
|
||
if serial in self._stop_requested:
|
||
_log.info(f"{serial} 用户已请求停止，不再重试 (job={job.name})")
|
||
return
|
||
_log.warning(f"{serial} 任务未成功(status={st})")
|
||
except DeviceOfflineError as e:
|
||
# 设备掉线，立即放弃，不重试
|
||
_log.error(f"{serial} 设备离线，放弃任务 {job.name}: {e}")
|
||
with self._lock:
|
||
self._running.pop(serial, None)
|
||
_update_status(serial, status="failed", task_job="",
|
||
last_error=f"设备离线: {e}")
|
||
return
|
||
except Exception as e:
|
||
_log.error(f"{serial} 执行异常: {e}", exc_info=True)
|
||
with self._lock:
|
||
self._running.pop(serial, None)
|
||
_update_status(serial, task_job="")
|
||
|
||
if attempt < max_attempts:
|
||
# 用户在 sleep 期间点停止也能中断
|
||
with self._lock:
|
||
if serial in self._stop_requested:
|
||
_log.info(f"{serial} 用户已请求停止，取消重试 (job={job.name})")
|
||
return
|
||
# 检查是否是端口耗尽类临时错误，需要更长退避等端口释放
|
||
with _WORKERS_LOCK:
|
||
err = _WORKERS.get(serial, {}).get("last_error", "")
|
||
if err.startswith("[transient]"):
|
||
# Windows TCP 端口耗尽，TIME_WAIT 默认 2-4 分钟，等 120 秒
|
||
extra_delay = max(delay, 120)
|
||
_log.info(f"{serial} ADB 连接临时错误（端口耗尽），{extra_delay}s 后重试 ({attempt+1}/{max_attempts})")
|
||
time.sleep(extra_delay)
|
||
else:
|
||
_log.info(f"{serial} {delay}s 后重试 ({attempt+1}/{max_attempts})")
|
||
time.sleep(delay)
|
||
|
||
_log.error(f"{serial} 任务 {job.name} 重试耗尽，放弃")
|
||
_update_status(serial, status="failed", last_error=f"{job.name} 重试{max_attempts}次失败")
|
||
finally:
|
||
# 清除停止标志：整个重试循环结束（成功/失败/停止）后允许下次任务
|
||
with self._lock:
|
||
self._stop_requested.discard(serial)
|
||
# 归还：本任务是抢占任务，结束后自动重新启动被抢占的原任务
|
||
if preempted_job:
|
||
j = self.jobs.get(preempted_job)
|
||
if j and j.enabled:
|
||
_log.info(f"{serial} 抢占任务 {job.name} 结束，归还设备给任务 {j.name}")
|
||
threading.Thread(target=self._run_job, args=(j,), daemon=True).start()
|
||
|
||
# ---- 运行控制 ----
|
||
def stop_device(self, serial):
|
||
"""停止指定设备的 worker，并阻止后续重试。
|
||
|
||
无论 worker 当前在运行还是在重试 sleep 中，都会阻止下一次重试。
|
||
"""
|
||
with self._lock:
|
||
self._stop_requested.add(serial)
|
||
info = self._running.get(serial)
|
||
if info:
|
||
w = info.get("worker")
|
||
if w and w.is_alive():
|
||
w.stop()
|
||
return True
|
||
# worker 已结束但重试循环可能还在 sleep —— 仍然返回 True 表示已阻止重试
|
||
return True
|
||
|
||
def stop_all(self):
|
||
"""停止所有运行中的 worker，并阻止后续重试。"""
|
||
with self._lock:
|
||
items = list(self._running.items())
|
||
for s, _ in items:
|
||
self._stop_requested.add(s)
|
||
stopped = []
|
||
for serial, info in items:
|
||
w = info.get("worker")
|
||
if w and w.is_alive():
|
||
w.stop()
|
||
stopped.append(serial)
|
||
return stopped
|
||
|
||
def get_running(self):
|
||
with self._lock:
|
||
return {s: dict(v) for s, v in self._running.items()}
|
||
|
||
# ---- 状态查询（带缓存，避免设备列表查询阻塞前端）----
|
||
_status_cache = None # (timestamp, data, error)
|
||
_status_cache_lock = threading.Lock()
|
||
_STATUS_CACHE_TTL = 5.0 # 缓存 5 秒，前端 5 秒刷新刚好命中
|
||
|
||
def get_status(self):
|
||
"""综合状态：设备池（清单+在线）+ 本地 worker + 运行中的任务。
|
||
|
||
带 5 秒缓存：避免每次 /api/status 都查库/adb 阻塞 Flask。
|
||
worker 状态实时读（内存，无 IO），不受缓存影响。
|
||
"""
|
||
with self._status_cache_lock:
|
||
if self._status_cache:
|
||
ts, cached, err = self._status_cache
|
||
if time.time() - ts < self._STATUS_CACHE_TTL:
|
||
if err:
|
||
return None, err
|
||
# 用缓存的设备列表 + 实时 worker 状态重新组装
|
||
return self._merge_status(cached), None
|
||
# 缓存过期或不存在，重新拉设备池
|
||
try:
|
||
configured = device_pool.list_configured()
|
||
except Exception as e:
|
||
with self._status_cache_lock:
|
||
self._status_cache = (time.time(), None, str(e))
|
||
return None, f"获取设备列表失败: {e}"
|
||
with self._status_cache_lock:
|
||
self._status_cache = (time.time(), configured, None)
|
||
return self._merge_status(configured), None
|
||
|
||
def _merge_status(self, configured_serials):
|
||
"""用设备池清单 + 实时 worker 状态 + 前台 App 组装返回结果。"""
|
||
worker_status = {w["serial"]: w for w in get_all_worker_status()}
|
||
running = self.get_running()
|
||
try:
|
||
online = set(device_pool.list_online())
|
||
except Exception:
|
||
online = set()
|
||
try:
|
||
names = {d["serial"]: d["name"] for d in device_pool.list_devices()}
|
||
except Exception:
|
||
names = {}
|
||
|
||
# 清理陈旧状态：serial 已不在设备池、且没有在跑 worker 的条目，
|
||
# 避免设备被删除后其失败记录仍残留在"异常汇总"里
|
||
configured = set(configured_serials)
|
||
for serial, w in list(worker_status.items()):
|
||
if serial in configured or serial in running:
|
||
continue
|
||
if w.get("status") in ("running", "connecting"):
|
||
continue
|
||
_remove_worker(serial)
|
||
|
||
result = []
|
||
for serial in configured_serials:
|
||
is_online = serial in online
|
||
w = worker_status.get(serial, {})
|
||
r = running.get(serial, {})
|
||
result.append({
|
||
"serial": serial,
|
||
"model": w.get("model", ""),
|
||
"device_name": names.get(serial, ""),
|
||
"present": is_online,
|
||
"ready": is_online, # 阶段 1：ready 概念并入在线状态
|
||
"stf_occupied": False, # 阶段 1：已无 STF 占用（阶段 3 删字段）
|
||
"owner": "",
|
||
"worker_status": w.get("status", "idle"),
|
||
"foreground_app": self._fg_scanner.get(serial),
|
||
# 通用进度字段（任意 app 通用，前端统一解析展示）
|
||
# 结构：{"done": int, "total": int, "unit": str, "action_counts": dict}
|
||
"progress": w.get("progress", {}),
|
||
"current_action": w.get("current_action", ""),
|
||
"last_error": w.get("last_error", ""),
|
||
"last_warning": w.get("last_warning", ""),
|
||
"running_job": r.get("job_id", ""),
|
||
"task_job": w.get("task_job", ""),
|
||
"attempt": r.get("attempt", 0),
|
||
"end_time": w.get("end_time", 0),
|
||
})
|
||
return result
|
||
|
||
def list_all_serials(self):
|
||
"""返回设备池在线设备的 serial 列表（供分组表单勾选用）。"""
|
||
try:
|
||
return device_pool.list_ready()
|
||
except Exception:
|
||
return []
|
||
|
||
def shutdown(self):
|
||
self.stop_all()
|
||
self._fg_scanner.stop()
|
||
stop_watchdog()
|
||
self.scheduler.shutdown(wait=False)
|