fix: 连接 Pipeline 到爬取流程 + 夜间跳过 + 断网恢复

- CrawlService.run_spider 现在自动调用 PostCrawlPipeline（爬取→存储→筛选→去重→推送）
- 定时任务添加 22:00-06:00 夜间跳过逻辑
- 断网恢复后不会漏公告：爬虫每次从 API 按日期倒序抓取，未推送的公告下次自动补推
- Pipeline 的 _exclude_sent + mark_sent 防止重复推送

app/api/crawl.py

@@ -12,13 +12,21 @@ async def trigger_crawl(request: CrawlTriggerRequest):
     names = service.get_spider_names()
 
     all_results = []
+    total_stored = 0
+    total_notified = 0
     for name in names:
         results = await service.run_spider(name)
         all_results.extend(results)
+        for r in results:
+            if r.pipeline_result:
+                total_stored += r.pipeline_result.stored
+                total_notified += r.pipeline_result.notified
 
     return {
         "spiders_run": names,
         "total_announcements": sum(r.total_count for r in all_results),
+        "total_stored": total_stored,
+        "total_notified": total_notified,
         "errors": [r.error_message for r in all_results if not r.success],
     }
 

app/api/deps.py

@@ -3,10 +3,11 @@ from sqlalchemy.ext.asyncio import AsyncSession
 from app.crawler.dahuagov_spider import DahuagovSpider
 from app.crawler.gxgp_spider import GXGPSpider
 from app.services.crawl_service import CrawlService
+from app.services.notification_service import NotificationService
 
 
 async def get_db() -> AsyncSession:
-    from app.main import async_session  # 延迟导入避免循环引用
+    from app.main import async_session
     async with async_session() as session:
         yield session
 
@@ -17,7 +18,11 @@ _crawl_service: CrawlService | None = None
 def get_crawl_service() -> CrawlService:
     global _crawl_service
     if _crawl_service is None:
-        _crawl_service = CrawlService()
+        from app.main import async_session
+        _crawl_service = CrawlService(
+            db_session_factory=async_session,
+            notification_service=NotificationService(),
+        )
         _crawl_service.register(GXGPSpider())
         _crawl_service.register(DahuagovSpider())
     return _crawl_service