refactor(backend): optimize database safety, validator performance, and scheduler concurrency

- Fix SQL injection risks in proxy_repo and task_repo
- Atomic acquire_pending with UPDATE ... RETURNING
- Reuse aiohttp ClientSession in ValidatorService
- Replace polling with asyncio.Event in SchedulerService
- Optimize ValidationQueue.drain with asyncio.Condition
- Concurrent plugin crawling with asyncio.gather
- Unify ProxyRaw model import path
- Fix test baseline and remove tracked __pycache__ files

app/core/plugin_system/base.py

@@ -1,20 +1,7 @@
 """插件基类 - 所有爬虫插件必须继承此基类"""
 from abc import ABC, abstractmethod
-from dataclasses import dataclass
 from typing import List, Dict, Any
-
-
-@dataclass
-class ProxyRaw:
-    """爬虫产出的原始代理数据"""
-    ip: str
-    port: int
-    protocol: str = "http"
-
-    def __post_init__(self):
-        self.protocol = self.protocol.lower().strip()
-        if self.protocol not in ("http", "https", "socks4", "socks5"):
-            self.protocol = "http"
+from app.models.domain import ProxyRaw
 
 
 class BaseCrawlerPlugin(ABC):

app/core/tasks/queue.py

@@ -40,6 +40,8 @@ class ValidationQueue:
         self._workers: list[asyncio.Task] = []
         self._running = False
         self._db_lock = asyncio.Lock()
+        self._pending_count = 0
+        self._condition = asyncio.Condition()
 
         # 统计
         self.valid_count = 0
@@ -58,6 +60,8 @@ class ValidationQueue:
             logger.info(f"ValidationQueue recovered {recovered} interrupted tasks")
         if pending:
             logger.info(f"ValidationQueue has {pending} pending tasks to process")
+        async with self._condition:
+            self._pending_count = pending
 
         for i in range(self.worker_count):
             self._workers.append(asyncio.create_task(self._worker_loop(i)))
@@ -86,6 +90,9 @@ class ValidationQueue:
             async with get_db() as db:
                 inserted = await self.task_repo.insert_batch(db, proxies)
         if inserted:
+            async with self._condition:
+                self._pending_count += inserted
+                self._condition.notify_all()
             for _ in range(min(inserted, self.worker_count)):
                 self._signal.put_nowait(None)
 
@@ -94,12 +101,9 @@ class ValidationQueue:
 
     async def drain(self):
         """等待队列中当前所有 pending 任务处理完毕"""
-        while True:
-            async with get_db() as db:
-                count = await self.task_repo.get_pending_count(db)
-            if count == 0:
-                break
-            await asyncio.sleep(0.5)
+        async with self._condition:
+            if self._pending_count > 0:
+                await self._condition.wait_for(lambda: self._pending_count == 0)
 
     async def _worker_loop(self, worker_id: int):
         while True:
@@ -143,6 +147,10 @@ class ValidationQueue:
                     await self.task_repo.complete_task(db, task["id"], False, 0.0)
                     self.invalid_count += 1
                     logger.debug(f"ValidationQueue: invalid {proxy.ip}:{proxy.port}")
+        async with self._condition:
+            self._pending_count = max(0, self._pending_count - 1)
+            if self._pending_count == 0:
+                self._condition.notify_all()
 
     def reset_stats(self):
         self.valid_count = 0