Files
ProxyPool/app/services/scheduler_service.py
祀梦 4ef7931941 fix: 全面修复代码问题并优化架构
修复问题:
- 添加缺失的 httpx 依赖到 requirements.txt
- 修复前端批量删除参数格式与后端不匹配(数组->对象数组)
- 移除 app/api/main.py 中重复创建 app 的冗余代码
- 修复 Plugins.vue v-model 直接修改 store 状态的 Vue 警告
- 修复 README 端口/启动命令文档与实际配置不一致
- 修正 pytest.ini 过时配置 (asyncio_default_fixture_loop_scope)
- 修复 WebUI index.html 语言设置为 zh-CN
- 修复 .gitignore 错误忽略 tests/ 目录

后端优化:
- 修复调度器默认间隔从 5 秒改为 30 分钟,避免无节制验证
- 修复 validate_all_now 在调度器停止时无法执行的 bug
- 设置保存后热更新运行中调度器的验证间隔
- 将 update_score 优化为原子单事务 SQL,消除并发竞态
- 导出功能改为真正的流式分批读取(iter_batches),降低大导出内存占用
- ProxyResponse Schema 补齐 response_time_ms 字段
- 日志级别改为从配置动态读取,不再硬编码 INFO
- 清理 validator_service 中的冗余 try/finally 代码

插件健壮性:
- 修复 ip3366/ip89/kuaidaili/proxylist_download/speedx/yundaili/proxyscrape
  的端口范围检查和 IPv6 地址解析问题(改用 rsplit + 1-65535 校验)
- 修复 PluginService.list_plugins 并发竞争条件
- 修复 run_all_plugins 去重逻辑与数据库 UNIQUE 约束保持一致
- 修复 proxyscrape 异常时错误跳过 fallback 的 bug

测试:
- 新增 7 个插件解析单元测试
- 新增 update_score 自动删除和 iter_batches 流式读取测试
- 全部 74 个测试通过
2026-04-04 21:03:43 +08:00

119 lines
4.5 KiB
Python

"""调度器服务 - 定时验证存量代理"""
import asyncio
from datetime import datetime
from app.core.db import get_db
from app.repositories.proxy_repo import ProxyRepository
from app.repositories.task_repo import ValidationTaskRepository
from app.core.tasks.queue import ValidationQueue
from app.core.config import settings as app_settings
from app.core.log import logger
from app.models.domain import ProxyRaw
from app.services.task_service import task_service
class SchedulerService:
"""代理验证调度器"""
def __init__(
self,
validation_queue: ValidationQueue,
proxy_repo: ProxyRepository = ProxyRepository(),
):
self.validation_queue = validation_queue
self.proxy_repo = proxy_repo
self.interval_minutes = 30
self.running = False
self._stop_event = asyncio.Event()
self._task: asyncio.Task | None = None
self._validate_task: asyncio.Task | None = None
async def start(self):
if self.running:
logger.warning("Scheduler already running")
return
self._stop_event.clear()
self.running = True
await self.validation_queue.start()
self._task = asyncio.create_task(self._run_loop())
logger.info("Scheduler started")
async def stop(self):
self.running = False
self._stop_event.set()
if self._task:
self._task.cancel()
try:
await self._task
except asyncio.CancelledError:
pass
self._task = None
await self.validation_queue.stop()
logger.info("Scheduler stopped")
def cancel_validate_task(self):
"""取消正在执行的全量验证后台任务"""
if self._validate_task and not self._validate_task.done():
self._validate_task.cancel()
async def validate_all_now(self):
"""立即执行一次全量验证(后台运行,不阻塞)"""
if self._validate_task and not self._validate_task.done():
return
self._validate_task = asyncio.create_task(self._do_validate_all(from_loop=False))
async def _run_loop(self):
"""定时循环"""
while self.running:
try:
# 清理过期任务,防止内存无限增长
task_service.cleanup_old_tasks()
await self._do_validate_all(from_loop=True)
except Exception as e:
logger.error(f"Scheduler loop error: {e}", exc_info=True)
# 等待下一次
try:
await asyncio.wait_for(self._stop_event.wait(), timeout=self.interval_minutes * 60)
except asyncio.TimeoutError:
pass
async def _do_validate_all(self, from_loop: bool = True):
"""验证数据库中所有存量代理"""
queue_started_here = False
try:
logger.info("Starting scheduled validation for all proxies")
# 如果队列未运行,临时启动它(适用于 validate_all_now 在调度器停止时调用)
if not self.validation_queue._running:
await self.validation_queue.start()
queue_started_here = True
async with get_db() as db:
# 清理 7 天前的验证任务记录,防止表无限增长
cleaned = await ValidationTaskRepository.cleanup_old(db, days=7)
if cleaned:
logger.info(f"Cleaned up {cleaned} old validation tasks")
proxies = await self.proxy_repo.list_all(db)
if not proxies:
logger.info("No proxies to validate")
return
logger.info(f"Validating {len(proxies)} proxies from database")
# 批量提交到验证队列,不再阻塞等待 drain
batch_size = 100
total_batches = (len(proxies) - 1) // batch_size + 1
for i in range(0, len(proxies), batch_size):
if from_loop and not self.running:
break
batch = proxies[i : i + batch_size]
await self.validation_queue.submit([
ProxyRaw(p.ip, p.port, p.protocol) for p in batch
])
logger.info(f"Submitted batch {i // batch_size + 1}/{total_batches}")
logger.info("Scheduled validation batches submitted")
except Exception as e:
logger.error(f"Scheduled validation error: {e}", exc_info=True)
finally:
if queue_started_here:
await self.validation_queue.stop()