fix: content quality, image format, task monitor, calendar data source, search UI & sort

2026-05-26 11:26:10 +08:00
parent b2d043b231
commit ac8644d752
36 changed files with 2294 additions and 873 deletions
@@ -27,6 +27,8 @@ MODULES = {
    "scheduled_optimize": {"name": "🔍 合规审查", "cron": "03:00"},
    "scheduled_optimize_sources": {"name": "📡 信息源优化", "cron": "05:00"},
    "scheduled_metrics_sync": {"name": "📊 指标同步", "cron": "06:00"},
+    "scheduled_reset_search_usage": {"name": "🔁 搜索用量重置", "cron": "00:05"},
+    "scheduled_task_monitor": {"name": "⏰ 任务监控", "cron": "*"},
 }

 def _log_task(module_id: str, status: str, message: str = None,
@@ -107,6 +109,7 @@ class TaskScheduler:
            ("scheduled_optimize", self._run_optimize, "合规审查"),
            ("scheduled_optimize_sources", self._run_optimize_sources, "信息源优化"),
            ("scheduled_metrics_sync", self._run_metrics_sync, "指标同步"),
+            ("scheduled_reset_search_usage", self._run_reset_search_usage, "搜索用量重置"),
        ]

        for module_id, fn, name in MODULE_JOBS:
@@ -129,6 +132,17 @@ class TaskScheduler:
            )
            logger.info(f"调度任务: {module_id} -> {schedule}")

+        # 每小时运行的任务监控：检测卡死/中断任务
+        self.scheduler.add_job(
+            self._run_task_monitor,
+            CronTrigger(hour='*/1'),
+            id='scheduled_task_monitor',
+            replace_existing=True,
+            max_instances=1,
+            coalesce=True
+        )
+        logger.info("调度任务: scheduled_task_monitor -> 每小时")
+
        self.scheduler.start()
        self._started = True
        logger.info("Scheduler started with dynamic schedule from TaskConfig")
@@ -441,6 +455,95 @@ class TaskScheduler:
                      started_at=started, finished_at=datetime.now(timezone.utc))
            logger.exception("[Scheduled] Metrics sync failed: %s", e)

+    def _run_reset_search_usage(self):
+        """每日凌晨重置搜索 API 提供商用量计数"""
+        started = datetime.now(timezone.utc)
+        _log_task("scheduled_reset_search_usage", "running", started_at=started)
+        try:
+            from ..database import SessionLocal
+            from ..models import SearchProvider
+            db = SessionLocal()
+            try:
+                total = db.query(SearchProvider).update({SearchProvider.usage_today: 0, SearchProvider.last_used_at: None})
+                db.commit()
+                _log_task("scheduled_reset_search_usage", "success",
+                          message=f"已重置 {total} 个提供商用量",
+                          result_data={"reset_count": total},
+                          started_at=started, finished_at=datetime.now(timezone.utc))
+                logger.info("[Scheduled] Reset %d search providers usage", total)
+            finally:
+                db.close()
+        except Exception as e:
+            _log_task("scheduled_reset_search_usage", "failed",
+                      message=str(e),
+                      error_trace=traceback.format_exc(),
+                      started_at=started, finished_at=datetime.now(timezone.utc))
+            logger.exception("[Scheduled] Reset search usage failed: %s", e)
+
+    def _run_task_monitor(self):
+        """每小时检查卡死/中断的任务，标记为失败"""
+        started = datetime.now(timezone.utc)
+        _log_task("scheduled_task_monitor", "running", started_at=started)
+        stuck_tasklog_timeout = 7200  # 超过2小时视为卡死
+        stuck_contenttask_timeout = 10800  # 超过3小时视为卡死
+        try:
+            from ..database import SessionLocal
+            from ..models import TaskLog, ContentTask
+            db = SessionLocal()
+            try:
+                now = datetime.now(timezone.utc)
+                cutoff_tasklog = now.timestamp() - stuck_tasklog_timeout
+                cutoff_content = now.timestamp() - stuck_contenttask_timeout
+                marked = 0
+
+                # 检查 TaskLog 中卡死的 running 记录
+                stuck_logs = db.query(TaskLog).filter(
+                    TaskLog.status == "running",
+                    TaskLog.started_at.isnot(None)
+                ).all()
+                for log in stuck_logs:
+                    if log.started_at.timestamp() < cutoff_tasklog:
+                        log.status = "failed"
+                        log.finished_at = now
+                        log.error_trace = "系统监控：任务运行超时（超过2小时）或进程中断，已自动标记为失败"
+                        if log.started_at:
+                            log.duration = int((now - log.started_at).total_seconds())
+                        marked += 1
+                        logger.warning("[TaskMonitor] 标记 TaskLog %d (%s) 为失败（超时）", log.id, log.module_id)
+
+                # 检查 ContentTask 中卡死的 running 记录
+                stuck_tasks = db.query(ContentTask).filter(
+                    ContentTask.status == "running",
+                    ContentTask.started_at.isnot(None)
+                ).all()
+                for task in stuck_tasks:
+                    if task.started_at.timestamp() < cutoff_content:
+                        task.status = "failed"
+                        task.finished_at = now
+                        task.error_msg = "系统监控：任务运行超时（超过3小时）或进程中断，已自动标记为失败"
+                        if task.started_at:
+                            task.duration = int((now - task.started_at).total_seconds())
+                        marked += 1
+                        logger.warning("[TaskMonitor] 标记 ContentTask %s (%s) 为失败（超时）", task.task_id, task.stage)
+
+                if marked:
+                    db.commit()
+                    logger.info("[TaskMonitor] 已标记 %d 个卡死任务为失败", marked)
+
+                _log_task("scheduled_task_monitor", "success",
+                          message=f"检查完成，标记 {marked} 个卡死任务",
+                          result_data={"marked_failed": marked},
+                          started_at=started, finished_at=datetime.now(timezone.utc))
+            finally:
+                db.close()
+        except Exception as e:
+            import traceback
+            _log_task("scheduled_task_monitor", "failed",
+                      message=str(e),
+                      error_trace=traceback.format_exc(),
+                      started_at=started, finished_at=datetime.now(timezone.utc))
+            logger.exception("[TaskMonitor] 监控检查失败: %s", e)
+
    def get_jobs(self):
        """返回当前所有定时任务的状态"""
        jobs = []