perf: 全栈查询优化、前端缓存去重与页面可见性优化

后端:
- SQL count 查询统一改用 func.count() 子查询替代 query.count()
- Dashboard/Audit 等页面多次独立查询合并为单次聚合查询
- Provider summary 列表改为批量查询消除 N+1 问题
- DailyStats 逐天循环查询改为 CASE 分桶单次查询
- 使用 load_only() 减少不必要的列加载
- cache_decorator 支持嵌套属性路径解析(dotted vary_by)
- 多个管理/公共端点新增 @cache_result 缓存装饰

前端:
- cache.ts 新增 in-flight 请求复用、dedupedRequest、buildCacheKey
- 大量 API 调用添加前端缓存或去重
- 多个页面定时器在标签页隐藏时暂停、可见时恢复
- Auth 检查从 setInterval 改为 storage + visibilitychange 事件驱动
- 请求竞态防护(requestId 模式)

数据库:
- Usage 表新增 idx_usage_status_user_created 复合索引

src/services/management_token/service.py

@@ -6,6 +6,7 @@ import ipaddress
 from datetime import datetime, timezone
 from typing import Any
 
+from sqlalchemy import func
 from sqlalchemy.exc import IntegrityError
 from sqlalchemy.orm import Session
 
@@ -159,7 +160,12 @@ class ManagementTokenService:
             ValueError: 如果名称已存在或超过数量限制
         """
         # 检查用户 Token 数量限制
-        token_count = db.query(ManagementToken).filter(ManagementToken.user_id == user_id).count()
+        token_count = int(
+            db.query(func.count(ManagementToken.id))
+            .filter(ManagementToken.user_id == user_id)
+            .scalar()
+            or 0
+        )
         max_tokens = config.management_token_max_per_user
         if token_count >= max_tokens:
             raise ValueError(f"已达到 Token 数量上限（{max_tokens}）")
@@ -245,7 +251,7 @@ class ManagementTokenService:
         if is_active is not None:
             query = query.filter(ManagementToken.is_active == is_active)
 
-        total = query.count()
+        total = int(query.with_entities(func.count(ManagementToken.id)).scalar() or 0)
         tokens = query.order_by(ManagementToken.created_at.desc()).offset(skip).limit(limit).all()
 
         return tokens, total

src/services/model/service.py

@@ -6,7 +6,7 @@ from __future__ import annotations
 
 import asyncio
 
-from sqlalchemy import and_
+from sqlalchemy import and_, func
 from sqlalchemy.exc import IntegrityError
 from sqlalchemy.orm import Session
 
@@ -255,14 +255,15 @@ class ModelService:
 
         # 检查这是否是该 GlobalModel 的最后一个关联提供商
         if model.global_model_id:
-            other_implementations = (
-                db.query(Model)
+            other_implementations = int(
+                db.query(func.count(Model.id))
                 .filter(
                     Model.global_model_id == model.global_model_id,
                     Model.id != model_id,
                     Model.is_active == True,
                 )
-                .count()
+                .scalar()
+                or 0
             )
 
             if other_implementations == 0:

src/services/proxy_node/service.py

@@ -14,7 +14,7 @@ from typing import Any
 from urllib.parse import urlparse
 
 import httpx
-from sqlalchemy import update
+from sqlalchemy import func, update
 from sqlalchemy.orm import Session
 
 from src.core.exceptions import InvalidRequestException, NotFoundException
@@ -452,7 +452,7 @@ class ProxyNodeService:
                 raise InvalidRequestException(f"status 必须是以下之一: {sorted(allowed)}", "status")
             query = query.filter(ProxyNode.status == ProxyNodeStatus(normalized))
 
-        total = query.count()
+        total = int(query.with_entities(func.count(ProxyNode.id)).scalar() or 0)
         nodes = query.order_by(ProxyNode.name.asc()).offset(skip).limit(limit).all()
         return nodes, total
 

src/services/system/announcement.py

@@ -6,7 +6,7 @@ from __future__ import annotations
 
 from datetime import datetime, timezone
 
-from sqlalchemy import or_
+from sqlalchemy import func, or_
 from sqlalchemy.orm import Session
 
 from src.core.exceptions import ForbiddenException, NotFoundException
@@ -87,7 +87,7 @@ class AnnouncementService:
         )
 
         # 分页
-        total = query.count()
+        total = int(query.with_entities(func.count(Announcement.id)).scalar() or 0)
         announcements = query.offset(offset).limit(limit).all()
 
         # 获取已读状态

src/services/system/audit.py

@@ -338,8 +338,8 @@ class AuditService:
         )
 
         # 获取最近的可疑活动
-        recent_suspicious = (
-            db.query(AuditLog)
+        recent_suspicious = int(
+            db.query(func.count(AuditLog.id))
             .filter(
                 AuditLog.user_id == user_id,
                 AuditLog.event_type.in_(
@@ -350,7 +350,8 @@ class AuditService:
                 ),
                 AuditLog.created_at >= cutoff_time,
             )
-            .count()
+            .scalar()
+            or 0
         )
 
         return {

src/services/system/config.py

@@ -116,8 +116,8 @@ class SystemConfigService:
             "description": "详细日志保留天数，超过此天数后压缩 request_body 和 response_body 到压缩字段",
         },
         "compressed_log_retention_days": {
-            "value": 90,
-            "description": "压缩日志保留天数，超过此天数后删除压缩的 body 字段（保留headers和统计）",
+            "value": 30,
+            "description": "压缩记录保留天数，超过此天数后删除压缩的 body 字段（保留headers和统计）",
         },
         "header_retention_days": {
             "value": 90,
@@ -125,7 +125,7 @@ class SystemConfigService:
         },
         "log_retention_days": {
             "value": 365,
-            "description": "完整日志保留天数，超过此天数后删除整条记录（保留核心统计）",
+            "description": "请求记录保存天数，超过此天数后删除整条使用记录",
         },
         "enable_auto_cleanup": {
             "value": True,

src/services/system/maintenance_scheduler.py

@@ -1052,7 +1052,7 @@ class MaintenanceScheduler:
             # 获取配置参数
             detail_retention = SystemConfigService.get_config(db, "detail_log_retention_days", 7)
             compressed_retention = SystemConfigService.get_config(
-                db, "compressed_log_retention_days", 90
+                db, "compressed_log_retention_days", 30
             )
             header_retention = SystemConfigService.get_config(db, "header_retention_days", 90)
             log_retention = SystemConfigService.get_config(db, "log_retention_days", 365)

src/services/usage/active_requests.py

@@ -3,6 +3,7 @@ from __future__ import annotations
 from datetime import datetime, timedelta, timezone
 from typing import Any
 
+from sqlalchemy import func
 from sqlalchemy.orm import Session
 
 from src.core.api_format.metadata import can_passthrough_endpoint
@@ -104,13 +105,14 @@ class UsageActiveRequestsMixin:
         """
         cutoff_time = datetime.now(timezone.utc) - timedelta(minutes=timeout_minutes)
 
-        return (
-            db.query(Usage)
+        return int(
+            db.query(func.count(Usage.id))
             .filter(
                 Usage.status.in_(["pending", "streaming"]),
                 Usage.created_at < cutoff_time,
             )
-            .count()
+            .scalar()
+            or 0
         )
 
     @classmethod

src/services/usage/query.py

@@ -3,7 +3,7 @@ from __future__ import annotations
 from datetime import datetime, timedelta, timezone
 from typing import Any
 
-from sqlalchemy import func
+from sqlalchemy import case, func
 from sqlalchemy.orm import Session
 
 from src.core.logger import logger
@@ -239,6 +239,21 @@ class UsageQueryMixin:
             func.sum(Usage.total_tokens).label("total_tokens"),
             func.sum(Usage.total_cost_usd).label("total_cost_usd"),
             func.avg(Usage.response_time_ms).label("avg_response_time"),
+            func.sum(
+                case(
+                    (
+                        (Usage.status_code == 200) & Usage.response_time_ms.isnot(None),
+                        Usage.response_time_ms,
+                    ),
+                    else_=0,
+                )
+            ).label("success_response_time_sum"),
+            func.sum(
+                case(
+                    ((Usage.status_code == 200) & Usage.response_time_ms.isnot(None), 1),
+                    else_=0,
+                )
+            ).label("success_response_time_count"),
         )
 
         # 过滤掉 pending/streaming 状态的请求（与上方明细查询一致）
@@ -268,6 +283,8 @@ class UsageQueryMixin:
                 "avg_response_time_ms": (
                     float(row.avg_response_time) if row.avg_response_time else 0
                 ),
+                "success_response_time_sum_ms": float(row.success_response_time_sum or 0.0),
+                "success_response_time_count": int(row.success_response_time_count or 0),
             }
             for row in summary
         ]

src/services/user/service.py

@@ -266,7 +266,9 @@ class UserService:
         db.query(AnnouncementRead).filter(AnnouncementRead.user_id == user_id).delete(
             synchronize_session=False
         )
-        api_key_count = db.query(ApiKey).filter(ApiKey.user_id == user_id).count()
+        api_key_count = int(
+            db.query(func.count(ApiKey.id)).filter(ApiKey.user_id == user_id).scalar() or 0
+        )
         db.query(ApiKey).filter(ApiKey.user_id == user_id).delete(synchronize_session=False)
 
         # 现在删除用户（Usage, AuditLog, RequestAttempt 会通过数据库 SET NULL 保留）