Merge pull request #2 from fawney19/dev

Dev

README.md

@@ -61,6 +61,16 @@ python generate_keys.py  # 生成密钥, 并将生成的密钥填入 .env
 ./deploy.sh  # 自动构建、启动、迁移
 ```
 
+### 更新
+
+```bash
+# 拉取最新代码
+git pull
+
+# 自动部署脚本
+./deploy.sh
+```
+
 ### 本地开发
 
 ```bash

frontend/src/api/client.ts

@@ -1,6 +1,8 @@
-import axios from 'axios'
+import axios, { getAdapter } from 'axios'
-import type { AxiosInstance, AxiosRequestConfig, AxiosResponse } from 'axios'
+import type { AxiosInstance, AxiosRequestConfig, AxiosResponse, InternalAxiosRequestConfig, AxiosAdapter } from 'axios'
 import { NETWORK_CONFIG, AUTH_CONFIG } from '@/config/constants'
+import { isDemoMode } from '@/config/demo'
+import { handleMockRequest, setMockUserToken } from '@/mocks'
 import { log } from '@/utils/logger'
 
 // 在开发环境下使用代理,生产环境使用环境变量
@@ -42,6 +44,39 @@ function isRefreshableAuthError(errorDetail: string): boolean {
   return !nonRefreshableErrors.some((msg) => errorDetail.includes(msg))
 }
 
+/**
+ * 创建 Demo 模式的自定义 adapter
+ * 在 Demo 模式下拦截请求并返回 mock 数据
+ */
+function createDemoAdapter(defaultAdapter: AxiosAdapter) {
+  return async (config: InternalAxiosRequestConfig): Promise<AxiosResponse> => {
+    if (isDemoMode()) {
+      try {
+        const mockResponse = await handleMockRequest({
+          method: config.method?.toUpperCase(),
+          url: config.url,
+          data: config.data,
+          params: config.params,
+        })
+        if (mockResponse) {
+          // 确保响应包含 config
+          mockResponse.config = config
+          return mockResponse
+        }
+      } catch (error: any) {
+        // Mock 错误需要附加 config，否则 handleResponseError 会崩溃
+        if (error.response) {
+          error.config = config
+          error.response.config = config
+        }
+        throw error
+      }
+    }
+    // 非 Demo 模式或没有 mock 响应时，使用默认 adapter
+    return defaultAdapter(config)
+  }
+}
+
 class ApiClient {
   private client: AxiosInstance
   private token: string | null = null
@@ -57,6 +92,10 @@ class ApiClient {
       },
     })
 
+    // 设置自定义 adapter 处理 Demo 模式
+    const defaultAdapter = getAdapter(this.client.defaults.adapter)
+    this.client.defaults.adapter = createDemoAdapter(defaultAdapter)
+
     this.setupInterceptors()
   }
 
@@ -64,7 +103,7 @@ class ApiClient {
    * 配置请求和响应拦截器
    */
   private setupInterceptors(): void {
-    // 请求拦截器
+    // 请求拦截器 - 仅处理认证
     this.client.interceptors.request.use(
       (config) => {
         const requiresAuth = !isPublicEndpoint(config.url, config.method) &&
@@ -207,11 +246,19 @@ class ApiClient {
   setToken(token: string): void {
     this.token = token
     localStorage.setItem('access_token', token)
+    // 同步到 mock handler
+    if (isDemoMode()) {
+      setMockUserToken(token)
+    }
   }
 
   getToken(): string | null {
     if (!this.token) {
       this.token = localStorage.getItem('access_token')
+      // 页面刷新时，从 localStorage 恢复 token 到 mock handler
+      if (this.token && isDemoMode()) {
+        setMockUserToken(this.token)
+      }
     }
     return this.token
   }
@@ -220,12 +267,18 @@ class ApiClient {
     this.token = null
     localStorage.removeItem('access_token')
     localStorage.removeItem('refresh_token')
+    // 同步清除 mock token
+    if (isDemoMode()) {
+      setMockUserToken(null)
+    }
   }
 
   async refreshToken(refreshToken: string): Promise<AxiosResponse> {
+    // refreshToken 会通过 adapter 处理 Demo 模式
     return this.client.post('/api/auth/refresh', { refresh_token: refreshToken })
   }
 
+  // 以下方法直接委托给 axios client，Demo 模式由 adapter 统一处理
   async request<T = any>(config: AxiosRequestConfig): Promise<AxiosResponse<T>> {
     return this.client.request<T>(config)
   }

frontend/src/api/endpoints/types.ts

@@ -81,6 +81,7 @@ export interface EndpointHealthDetail {
   api_format: string
   health_score: number
   is_active: boolean
+  active_keys?: number
 }
 
 export interface EndpointHealthEvent {

frontend/src/config/demo.ts

@@ -0,0 +1,37 @@
+/**
+ * Demo Mode Configuration
+ * 用于 GitHub Pages 等静态托管环境的演示模式
+ */
+
+// 检测是否为演示模式环境
+export function isDemoMode(): boolean {
+  const hostname = window.location.hostname
+  return (
+    hostname.includes('github.io') ||
+    hostname.includes('vercel.app') ||
+    hostname.includes('netlify.app') ||
+    hostname.includes('pages.dev') ||
+    import.meta.env.VITE_DEMO_MODE === 'true'
+  )
+}
+
+// Demo 账号配置
+export const DEMO_ACCOUNTS = {
+  admin: {
+    email: 'admin@demo.aether.io',
+    password: 'demo123',
+    hint: '管理员账号'
+  },
+  user: {
+    email: 'user@demo.aether.io',
+    password: 'demo123',
+    hint: '普通用户'
+  }
+} as const
+
+// Demo 模式提示信息
+export const DEMO_MODE_INFO = {
+  title: '演示模式',
+  description: '当前处于演示模式，所有数据均为模拟数据，不会产生实际调用。',
+  accountHint: '可使用以下演示账号登录：'
+} as const

frontend/src/features/auth/components/LoginDialog.vue

@@ -11,6 +11,43 @@
         </h2>
       </div>
 
+      <!-- Demo 模式提示 -->
+      <div v-if="isDemo" class="rounded-lg border border-primary/20 dark:border-primary/30 bg-primary/5 dark:bg-primary/10 p-4">
+        <div class="flex items-start gap-3">
+          <div class="flex-shrink-0 text-primary dark:text-primary/90">
+            <svg class="h-5 w-5" viewBox="0 0 20 20" fill="currentColor">
+              <path fill-rule="evenodd" d="M18 10a8 8 0 11-16 0 8 8 0 0116 0zm-7-4a1 1 0 11-2 0 1 1 0 012 0zM9 9a.75.75 0 000 1.5h.253a.25.25 0 01.244.304l-.459 2.066A1.75 1.75 0 0010.747 15H11a.75.75 0 000-1.5h-.253a.25.25 0 01-.244-.304l.459-2.066A1.75 1.75 0 009.253 9H9z" clip-rule="evenodd" />
+            </svg>
+          </div>
+          <div class="flex-1 min-w-0">
+            <p class="text-sm font-medium text-foreground">
+              演示模式
+            </p>
+            <p class="mt-1 text-xs text-muted-foreground">
+              当前处于演示模式，所有数据均为模拟数据。
+            </p>
+            <div class="mt-3 space-y-2">
+              <button
+                type="button"
+                @click="fillDemoAccount('admin')"
+                class="flex items-center gap-2 text-xs text-muted-foreground hover:text-foreground transition-colors group"
+              >
+                <span class="inline-flex items-center justify-center w-5 h-5 rounded bg-primary/20 dark:bg-primary/30 text-primary text-[10px] font-bold group-hover:bg-primary/30 dark:group-hover:bg-primary/40 transition-colors">A</span>
+                <span>管理员：admin@demo.aether.io / demo123</span>
+              </button>
+              <button
+                type="button"
+                @click="fillDemoAccount('user')"
+                class="flex items-center gap-2 text-xs text-muted-foreground hover:text-foreground transition-colors group"
+              >
+                <span class="inline-flex items-center justify-center w-5 h-5 rounded bg-muted text-muted-foreground text-[10px] font-bold group-hover:bg-muted/80 transition-colors">U</span>
+                <span>普通用户：user@demo.aether.io / demo123</span>
+              </button>
+            </div>
+          </div>
+        </div>
+      </div>
+
       <!-- 登录表单 -->
       <form @submit.prevent="handleLogin" class="space-y-4">
         <div class="space-y-2">
@@ -32,14 +69,14 @@
             v-model="form.password"
             type="password"
             required
-            placeholder="••••••••"
+            placeholder="********"
             autocomplete="off"
             @keyup.enter="handleLogin"
           />
         </div>
 
         <!-- 提示信息 -->
-        <p class="text-xs text-slate-400 dark:text-muted-foreground/80">
+        <p v-if="!isDemo" class="text-xs text-slate-400 dark:text-muted-foreground/80">
           如需开通账户，请联系管理员配置访问权限
         </p>
       </form>
@@ -66,7 +103,7 @@
 </template>
 
 <script setup lang="ts">
-import { ref, watch } from 'vue'
+import { ref, watch, computed } from 'vue'
 import { useRouter } from 'vue-router'
 import { Dialog } from '@/components/ui'
 import Button from '@/components/ui/button.vue'
@@ -74,6 +111,7 @@ import Input from '@/components/ui/input.vue'
 import Label from '@/components/ui/label.vue'
 import { useAuthStore } from '@/stores/auth'
 import { useToast } from '@/composables/useToast'
+import { isDemoMode, DEMO_ACCOUNTS } from '@/config/demo'
 
 const props = defineProps<{
   modelValue: boolean
@@ -88,6 +126,7 @@ const authStore = useAuthStore()
 const { success: showSuccess, warning: showWarning, error: showError } = useToast()
 
 const isOpen = ref(props.modelValue)
+const isDemo = computed(() => isDemoMode())
 
 watch(() => props.modelValue, (val) => {
   isOpen.value = val
@@ -109,6 +148,12 @@ const form = ref({
   password: ''
 })
 
+function fillDemoAccount(type: 'admin' | 'user') {
+  const account = DEMO_ACCOUNTS[type]
+  form.value.email = account.email
+  form.value.password = account.password
+}
+
 async function handleLogin() {
   if (!form.value.email || !form.value.password) {
     showWarning('请输入邮箱和密码')

frontend/src/features/models/components/ModelDetailDrawer.vue

@@ -21,19 +21,20 @@
                     {{ model.is_active ? '活跃' : '停用' }}
                   </Badge>
                 </div>
-                <div class="flex items-center gap-2">
+                <div class="flex items-center gap-2 text-sm text-muted-foreground min-w-0">
-                  <span class="text-sm text-muted-foreground font-mono">{{ model.name }}</span>
+                  <span class="font-mono shrink-0">{{ model.name }}</span>
                   <button
-                    class="p-0.5 rounded hover:bg-muted transition-colors"
+                    class="p-0.5 rounded hover:bg-muted transition-colors shrink-0"
                     title="复制模型 ID"
                     @click="copyToClipboard(model.name)"
                   >
-                    <Copy class="w-3 h-3 text-muted-foreground" />
+                    <Copy class="w-3 h-3" />
                   </button>
+                  <template v-if="model.description">
+                    <span class="shrink-0">·</span>
+                    <span class="text-xs truncate" :title="model.description">{{ model.description }}</span>
+                  </template>
                 </div>
-                <p v-if="model.description" class="text-xs text-muted-foreground">
-                  {{ model.description }}
-                </p>
               </div>
               <div class="flex items-center gap-1 shrink-0">
                 <Button variant="ghost" size="icon" @click="$emit('edit-model', model)" title="编辑模型">

frontend/src/layouts/MainLayout.vue

@@ -101,6 +101,12 @@
              </div>
           </div>
 
+          <!-- Demo Mode Badge (center) -->
+          <div v-if="isDemo" class="flex items-center gap-2 px-3 py-1.5 rounded-full bg-amber-100 dark:bg-amber-900/30 text-amber-700 dark:text-amber-400 text-xs font-medium">
+            <AlertTriangle class="w-3.5 h-3.5" />
+            <span>演示模式</span>
+          </div>
+
           <div class="flex items-center gap-2">
                <!-- Theme Toggle -->
                <button
@@ -125,6 +131,7 @@ import { computed, ref, onMounted, onUnmounted } from 'vue'
 import { useRoute, useRouter } from 'vue-router'
 import { useAuthStore } from '@/stores/auth'
 import { useDarkMode } from '@/composables/useDarkMode'
+import { isDemoMode } from '@/config/demo'
 import Button from '@/components/ui/button.vue'
 import AppShell from '@/components/layout/AppShell.vue'
 import SidebarNav from '@/components/layout/SidebarNav.vue'
@@ -157,6 +164,7 @@ const router = useRouter()
 const route = useRoute()
 const authStore = useAuthStore()
 const { isDark, themeMode, toggleDarkMode } = useDarkMode()
+const isDemo = computed(() => isDemoMode())
 
 const showAuthError = ref(false)
 let authCheckInterval: number | null = null

frontend/src/mocks/data.ts

@@ -0,0 +1,868 @@
+/**
+ * Demo Mode Mock Data
+ * 演示模式的模拟数据
+ */
+
+import type { User, LoginResponse } from '@/api/auth'
+import type { DashboardStatsResponse, RecentRequest, ProviderStatus, DailyStatsResponse } from '@/api/dashboard'
+import type { User as AdminUser, ApiKey } from '@/api/users'
+import type { AdminApiKey, AdminApiKeysResponse } from '@/api/admin'
+import type { Profile, UsageResponse } from '@/api/me'
+import type { ProviderWithEndpointsSummary, GlobalModelResponse } from '@/api/endpoints/types'
+
+// ========== 用户数据 ==========
+
+export const MOCK_ADMIN_USER: User = {
+  id: 'demo-admin-uuid-0001',
+  username: 'Demo Admin',
+  email: 'admin@demo.aether.io',
+  role: 'admin',
+  is_active: true,
+  quota_usd: null,
+  used_usd: 156.78,
+  total_usd: 1234.56,
+  allowed_providers: null,
+  allowed_endpoints: null,
+  allowed_models: null,
+  created_at: '2024-01-01T00:00:00Z',
+  last_login_at: new Date().toISOString()
+}
+
+export const MOCK_NORMAL_USER: User = {
+  id: 'demo-user-uuid-0002',
+  username: 'Demo User',
+  email: 'user@demo.aether.io',
+  role: 'user',
+  is_active: true,
+  quota_usd: 100,
+  used_usd: 45.32,
+  total_usd: 245.32,
+  allowed_providers: null,
+  allowed_endpoints: null,
+  allowed_models: null,
+  created_at: '2024-06-01T00:00:00Z',
+  last_login_at: new Date().toISOString()
+}
+
+export const MOCK_LOGIN_RESPONSE_ADMIN: LoginResponse = {
+  access_token: 'demo-access-token-admin',
+  refresh_token: 'demo-refresh-token-admin',
+  token_type: 'bearer',
+  expires_in: 3600,
+  user_id: MOCK_ADMIN_USER.id,
+  email: MOCK_ADMIN_USER.email,
+  username: MOCK_ADMIN_USER.username,
+  role: 'admin'
+}
+
+export const MOCK_LOGIN_RESPONSE_USER: LoginResponse = {
+  access_token: 'demo-access-token-user',
+  refresh_token: 'demo-refresh-token-user',
+  token_type: 'bearer',
+  expires_in: 3600,
+  user_id: MOCK_NORMAL_USER.id,
+  email: MOCK_NORMAL_USER.email,
+  username: MOCK_NORMAL_USER.username,
+  role: 'user'
+}
+
+// ========== Profile 数据 ==========
+
+export const MOCK_ADMIN_PROFILE: Profile = {
+  id: MOCK_ADMIN_USER.id!,
+  email: MOCK_ADMIN_USER.email!,
+  username: MOCK_ADMIN_USER.username,
+  role: 'admin',
+  is_active: true,
+  quota_usd: null,
+  used_usd: 156.78,
+  total_usd: 1234.56,
+  created_at: '2024-01-01T00:00:00Z',
+  updated_at: new Date().toISOString(),
+  last_login_at: new Date().toISOString(),
+  preferences: {
+    theme: 'auto',
+    language: 'zh-CN'
+  }
+}
+
+export const MOCK_USER_PROFILE: Profile = {
+  id: MOCK_NORMAL_USER.id!,
+  email: MOCK_NORMAL_USER.email!,
+  username: MOCK_NORMAL_USER.username,
+  role: 'user',
+  is_active: true,
+  quota_usd: 100,
+  used_usd: 45.32,
+  total_usd: 245.32,
+  created_at: '2024-06-01T00:00:00Z',
+  updated_at: new Date().toISOString(),
+  last_login_at: new Date().toISOString(),
+  preferences: {
+    theme: 'auto',
+    language: 'zh-CN'
+  }
+}
+
+// ========== Dashboard 数据 ==========
+
+export const MOCK_DASHBOARD_STATS: DashboardStatsResponse = {
+  stats: [
+    {
+      name: '今日请求',
+      value: '1,234',
+      subValue: '成功率 99.2%',
+      change: '+12.5%',
+      changeType: 'increase',
+      icon: 'Activity'
+    },
+    {
+      name: '今日 Token',
+      value: '2.5M',
+      subValue: '输入 1.8M / 输出 0.7M',
+      change: '+8.3%',
+      changeType: 'increase',
+      icon: 'Zap'
+    },
+    {
+      name: '今日费用',
+      value: '$45.67',
+      subValue: '节省 $12.34 (21%)',
+      change: '-5.2%',
+      changeType: 'decrease',
+      icon: 'DollarSign'
+    },
+    {
+      name: '活跃用户',
+      value: '28',
+      subValue: '总用户 156',
+      change: '+3',
+      changeType: 'increase',
+      icon: 'Users'
+    }
+  ],
+  today: {
+    requests: 1234,
+    tokens: 2500000,
+    cost: 45.67,
+    actual_cost: 33.33,
+    cache_creation_tokens: 50000,
+    cache_read_tokens: 200000
+  },
+  api_keys: {
+    total: 45,
+    active: 38
+  },
+  tokens: {
+    month: 75000000
+  },
+  system_health: {
+    avg_response_time: 1.23,
+    error_rate: 0.8,
+    error_requests: 10,
+    fallback_count: 5,
+    total_requests: 1234
+  },
+  cost_stats: {
+    total_cost: 45.67,
+    total_actual_cost: 33.33,
+    cost_savings: 12.34
+  },
+  cache_stats: {
+    cache_creation_tokens: 50000,
+    cache_read_tokens: 200000,
+    cache_creation_cost: 0.25,
+    cache_read_cost: 0.10,
+    cache_hit_rate: 0.35,
+    total_cache_tokens: 250000
+  },
+  users: {
+    total: 156,
+    active: 28
+  },
+  token_breakdown: {
+    input: 1800000,
+    output: 700000,
+    cache_creation: 50000,
+    cache_read: 200000
+  }
+}
+
+export const MOCK_RECENT_REQUESTS: RecentRequest[] = [
+  { id: 'req-001', user: 'alice', model: 'claude-sonnet-4-20250514', tokens: 15234, time: '2 分钟前' },
+  { id: 'req-002', user: 'bob', model: 'gpt-4o', tokens: 8765, time: '5 分钟前' },
+  { id: 'req-003', user: 'charlie', model: 'claude-opus-4-20250514', tokens: 32100, time: '8 分钟前' },
+  { id: 'req-004', user: 'diana', model: 'gemini-2.0-flash', tokens: 4521, time: '12 分钟前' },
+  { id: 'req-005', user: 'eve', model: 'claude-sonnet-4-20250514', tokens: 9876, time: '15 分钟前' },
+  { id: 'req-006', user: 'frank', model: 'gpt-4o-mini', tokens: 2345, time: '18 分钟前' },
+  { id: 'req-007', user: 'grace', model: 'claude-haiku-3-5-20241022', tokens: 6789, time: '22 分钟前' },
+  { id: 'req-008', user: 'henry', model: 'gemini-2.5-pro', tokens: 12345, time: '25 分钟前' }
+]
+
+export const MOCK_PROVIDER_STATUS: ProviderStatus[] = [
+  { name: 'Anthropic Official', status: 'active', requests: 456 },
+  { name: 'OpenAI Official', status: 'active', requests: 389 },
+  { name: 'Google AI', status: 'active', requests: 234 },
+  { name: 'AWS Bedrock', status: 'active', requests: 89 },
+  { name: 'Azure OpenAI', status: 'inactive', requests: 0 },
+  { name: 'Vertex AI', status: 'active', requests: 66 }
+]
+
+// 生成过去7天的每日统计数据
+function generateDailyStats(): DailyStatsResponse {
+  const dailyStats = []
+  const now = new Date()
+
+  for (let i = 6; i >= 0; i--) {
+    const date = new Date(now)
+    date.setDate(date.getDate() - i)
+    const dateStr = date.toISOString().split('T')[0]
+
+    const baseRequests = 800 + Math.floor(Math.random() * 600)
+    const baseTokens = 1500000 + Math.floor(Math.random() * 1500000)
+    const baseCost = 30 + Math.random() * 30
+
+    dailyStats.push({
+      date: dateStr,
+      requests: baseRequests,
+      tokens: baseTokens,
+      cost: Number(baseCost.toFixed(2)),
+      avg_response_time: 0.8 + Math.random() * 0.8,
+      unique_models: 8 + Math.floor(Math.random() * 5),
+      unique_providers: 4 + Math.floor(Math.random() * 3),
+      model_breakdown: [
+        { model: 'claude-sonnet-4-20250514', requests: Math.floor(baseRequests * 0.35), tokens: Math.floor(baseTokens * 0.35), cost: Number((baseCost * 0.35).toFixed(2)) },
+        { model: 'gpt-4o', requests: Math.floor(baseRequests * 0.25), tokens: Math.floor(baseTokens * 0.25), cost: Number((baseCost * 0.25).toFixed(2)) },
+        { model: 'claude-opus-4-20250514', requests: Math.floor(baseRequests * 0.15), tokens: Math.floor(baseTokens * 0.15), cost: Number((baseCost * 0.20).toFixed(2)) },
+        { model: 'gemini-2.0-flash', requests: Math.floor(baseRequests * 0.15), tokens: Math.floor(baseTokens * 0.15), cost: Number((baseCost * 0.10).toFixed(2)) },
+        { model: 'claude-haiku-3-5-20241022', requests: Math.floor(baseRequests * 0.10), tokens: Math.floor(baseTokens * 0.10), cost: Number((baseCost * 0.10).toFixed(2)) }
+      ]
+    })
+  }
+
+  return {
+    daily_stats: dailyStats,
+    model_summary: [
+      { model: 'claude-sonnet-4-20250514', requests: 2456, tokens: 8500000, cost: 125.45, avg_response_time: 1.2, cost_per_request: 0.051, tokens_per_request: 3461 },
+      { model: 'gpt-4o', requests: 1823, tokens: 6200000, cost: 98.32, avg_response_time: 0.9, cost_per_request: 0.054, tokens_per_request: 3401 },
+      { model: 'claude-opus-4-20250514', requests: 987, tokens: 4100000, cost: 156.78, avg_response_time: 2.1, cost_per_request: 0.159, tokens_per_request: 4154 },
+      { model: 'gemini-2.0-flash', requests: 1234, tokens: 3800000, cost: 28.56, avg_response_time: 0.6, cost_per_request: 0.023, tokens_per_request: 3079 },
+      { model: 'claude-haiku-3-5-20241022', requests: 2100, tokens: 5200000, cost: 32.10, avg_response_time: 0.5, cost_per_request: 0.015, tokens_per_request: 2476 }
+    ],
+    period: {
+      start_date: dailyStats[0].date,
+      end_date: dailyStats[dailyStats.length - 1].date,
+      days: 7
+    }
+  }
+}
+
+export const MOCK_DAILY_STATS = generateDailyStats()
+
+// ========== 用户管理数据 ==========
+
+export const MOCK_ALL_USERS: AdminUser[] = [
+  {
+    id: 'demo-admin-uuid-0001',
+    username: 'Demo Admin',
+    email: 'admin@demo.aether.io',
+    role: 'admin',
+    is_active: true,
+    quota_usd: null,
+    used_usd: 156.78,
+    total_usd: 1234.56,
+    allowed_providers: null,
+    allowed_endpoints: null,
+    allowed_models: null,
+    created_at: '2024-01-01T00:00:00Z'
+  },
+  {
+    id: 'demo-user-uuid-0002',
+    username: 'Demo User',
+    email: 'user@demo.aether.io',
+    role: 'user',
+    is_active: true,
+    quota_usd: 100,
+    used_usd: 45.32,
+    total_usd: 245.32,
+    allowed_providers: null,
+    allowed_endpoints: null,
+    allowed_models: null,
+    created_at: '2024-06-01T00:00:00Z'
+  },
+  {
+    id: 'demo-user-uuid-0003',
+    username: 'Alice Wang',
+    email: 'alice@example.com',
+    role: 'user',
+    is_active: true,
+    quota_usd: 50,
+    used_usd: 23.45,
+    total_usd: 123.45,
+    allowed_providers: null,
+    allowed_endpoints: null,
+    allowed_models: null,
+    created_at: '2024-03-15T00:00:00Z'
+  },
+  {
+    id: 'demo-user-uuid-0004',
+    username: 'Bob Zhang',
+    email: 'bob@example.com',
+    role: 'user',
+    is_active: true,
+    quota_usd: 200,
+    used_usd: 89.12,
+    total_usd: 589.12,
+    allowed_providers: null,
+    allowed_endpoints: null,
+    allowed_models: null,
+    created_at: '2024-02-20T00:00:00Z'
+  },
+  {
+    id: 'demo-user-uuid-0005',
+    username: 'Charlie Li',
+    email: 'charlie@example.com',
+    role: 'user',
+    is_active: false,
+    quota_usd: 30,
+    used_usd: 30.00,
+    total_usd: 30.00,
+    allowed_providers: null,
+    allowed_endpoints: null,
+    allowed_models: null,
+    created_at: '2024-04-10T00:00:00Z'
+  }
+]
+
+// ========== API Key 数据 ==========
+
+export const MOCK_USER_API_KEYS: ApiKey[] = [
+  {
+    id: 'key-uuid-001',
+    key_display: 'sk-ae...x7f9',
+    name: '开发环境',
+    created_at: '2024-06-15T00:00:00Z',
+    last_used_at: new Date().toISOString(),
+    is_active: true,
+    is_standalone: false,
+    total_requests: 1234,
+    total_cost_usd: 45.67
+  },
+  {
+    id: 'key-uuid-002',
+    key_display: 'sk-ae...m2k8',
+    name: '生产环境',
+    created_at: '2024-07-01T00:00:00Z',
+    last_used_at: new Date().toISOString(),
+    is_active: true,
+    is_standalone: false,
+    total_requests: 5678,
+    total_cost_usd: 123.45
+  },
+  {
+    id: 'key-uuid-003',
+    key_display: 'sk-ae...p9q1',
+    name: '测试用途',
+    created_at: '2024-08-01T00:00:00Z',
+    is_active: false,
+    is_standalone: false,
+    total_requests: 100,
+    total_cost_usd: 2.34
+  }
+]
+
+export const MOCK_ADMIN_API_KEYS: AdminApiKeysResponse = {
+  api_keys: [
+    {
+      id: 'standalone-key-001',
+      user_id: 'demo-user-uuid-0002',
+      user_email: 'user@demo.aether.io',
+      username: 'Demo User',
+      name: '独立余额 Key #1',
+      key_display: 'sk-sa...abc1',
+      is_active: true,
+      is_standalone: true,
+      balance_used_usd: 25.50,
+      current_balance_usd: 74.50,
+      total_requests: 500,
+      total_tokens: 1500000,
+      total_cost_usd: 25.50,
+      created_at: '2024-09-01T00:00:00Z',
+      last_used_at: new Date().toISOString()
+    },
+    {
+      id: 'standalone-key-002',
+      user_id: 'demo-user-uuid-0003',
+      user_email: 'alice@example.com',
+      username: 'Alice Wang',
+      name: '独立余额 Key #2',
+      key_display: 'sk-sa...def2',
+      is_active: true,
+      is_standalone: true,
+      balance_used_usd: 45.00,
+      current_balance_usd: 55.00,
+      total_requests: 800,
+      total_tokens: 2400000,
+      total_cost_usd: 45.00,
+      rate_limit: 60,
+      created_at: '2024-08-15T00:00:00Z',
+      last_used_at: new Date().toISOString()
+    }
+  ],
+  total: 2,
+  limit: 20,
+  skip: 0
+}
+
+// ========== Provider 数据 ==========
+
+export const MOCK_PROVIDERS: ProviderWithEndpointsSummary[] = [
+  {
+    id: 'provider-001',
+    name: 'duck_coding_free',
+    display_name: 'DuckCodingFree',
+    description: '',
+    website: 'https://duckcoding.com',
+    provider_priority: 1,
+    billing_type: 'free_tier',
+    monthly_used_usd: 0.0,
+    is_active: true,
+    total_endpoints: 3,
+    active_endpoints: 3,
+    total_keys: 3,
+    active_keys: 3,
+    total_models: 7,
+    active_models: 7,
+    avg_health_score: 0.91,
+    unhealthy_endpoints: 0,
+    api_formats: ['CLAUDE_CLI', 'GEMINI_CLI', 'OPENAI_CLI'],
+    endpoint_health_details: [
+      { api_format: 'CLAUDE_CLI', health_score: 0.73, is_active: true, active_keys: 1 },
+      { api_format: 'GEMINI_CLI', health_score: 1.0, is_active: true, active_keys: 1 },
+      { api_format: 'OPENAI_CLI', health_score: 1.0, is_active: true, active_keys: 1 }
+    ],
+    created_at: '2024-12-09T14:10:36.446217+08:00',
+    updated_at: new Date().toISOString()
+  },
+  {
+    id: 'provider-002',
+    name: 'open_claude_code',
+    display_name: 'OpenClaudeCode',
+    description: '',
+    website: 'https://www.openclaudecode.cn',
+    provider_priority: 2,
+    billing_type: 'pay_as_you_go',
+    monthly_used_usd: 545.18,
+    is_active: true,
+    total_endpoints: 2,
+    active_endpoints: 2,
+    total_keys: 3,
+    active_keys: 3,
+    total_models: 3,
+    active_models: 1,
+    avg_health_score: 0.825,
+    unhealthy_endpoints: 0,
+    api_formats: ['CLAUDE', 'CLAUDE_CLI'],
+    endpoint_health_details: [
+      { api_format: 'CLAUDE', health_score: 1.0, is_active: true, active_keys: 2 },
+      { api_format: 'CLAUDE_CLI', health_score: 0.65, is_active: true, active_keys: 1 }
+    ],
+    created_at: '2024-12-07T22:58:15.044538+08:00',
+    updated_at: new Date().toISOString()
+  },
+  {
+    id: 'provider-003',
+    name: '88_code',
+    display_name: '88Code',
+    description: '',
+    website: 'https://www.88code.org/',
+    provider_priority: 3,
+    billing_type: 'pay_as_you_go',
+    monthly_used_usd: 33.36,
+    is_active: true,
+    total_endpoints: 2,
+    active_endpoints: 2,
+    total_keys: 2,
+    active_keys: 2,
+    total_models: 5,
+    active_models: 5,
+    avg_health_score: 1.0,
+    unhealthy_endpoints: 0,
+    api_formats: ['CLAUDE_CLI', 'OPENAI_CLI'],
+    endpoint_health_details: [
+      { api_format: 'CLAUDE_CLI', health_score: 1.0, is_active: true, active_keys: 1 },
+      { api_format: 'OPENAI_CLI', health_score: 1.0, is_active: true, active_keys: 1 }
+    ],
+    created_at: '2024-12-07T22:56:46.361092+08:00',
+    updated_at: new Date().toISOString()
+  },
+  {
+    id: 'provider-004',
+    name: 'ikun_code',
+    display_name: 'IKunCode',
+    description: '',
+    website: 'https://api.ikuncode.cc',
+    provider_priority: 4,
+    billing_type: 'pay_as_you_go',
+    monthly_used_usd: 268.65,
+    is_active: true,
+    total_endpoints: 4,
+    active_endpoints: 4,
+    total_keys: 3,
+    active_keys: 3,
+    total_models: 7,
+    active_models: 7,
+    avg_health_score: 1.0,
+    unhealthy_endpoints: 0,
+    api_formats: ['CLAUDE_CLI', 'GEMINI', 'GEMINI_CLI', 'OPENAI_CLI'],
+    endpoint_health_details: [
+      { api_format: 'CLAUDE_CLI', health_score: 1.0, is_active: true, active_keys: 1 },
+      { api_format: 'GEMINI', health_score: 1.0, is_active: true, active_keys: 1 },
+      { api_format: 'GEMINI_CLI', health_score: 1.0, is_active: true, active_keys: 1 },
+      { api_format: 'OPENAI_CLI', health_score: 1.0, is_active: true, active_keys: 1 }
+    ],
+    created_at: '2024-12-07T15:16:55.807595+08:00',
+    updated_at: new Date().toISOString()
+  },
+  {
+    id: 'provider-005',
+    name: 'duck_coding',
+    display_name: 'DuckCoding',
+    description: '',
+    website: 'https://duckcoding.com',
+    provider_priority: 5,
+    billing_type: 'pay_as_you_go',
+    monthly_used_usd: 5.29,
+    is_active: true,
+    total_endpoints: 6,
+    active_endpoints: 6,
+    total_keys: 11,
+    active_keys: 11,
+    total_models: 8,
+    active_models: 8,
+    avg_health_score: 0.863,
+    unhealthy_endpoints: 1,
+    api_formats: ['CLAUDE', 'CLAUDE_CLI', 'GEMINI', 'GEMINI_CLI', 'OPENAI', 'OPENAI_CLI'],
+    endpoint_health_details: [
+      { api_format: 'CLAUDE', health_score: 1.0, is_active: true, active_keys: 2 },
+      { api_format: 'CLAUDE_CLI', health_score: 0.48, is_active: true, active_keys: 2 },
+      { api_format: 'GEMINI', health_score: 1.0, is_active: true, active_keys: 2 },
+      { api_format: 'GEMINI_CLI', health_score: 0.85, is_active: true, active_keys: 2 },
+      { api_format: 'OPENAI', health_score: 0.85, is_active: true, active_keys: 2 },
+      { api_format: 'OPENAI_CLI', health_score: 1.0, is_active: true, active_keys: 1 }
+    ],
+    created_at: '2024-12-07T22:56:09.712806+08:00',
+    updated_at: new Date().toISOString()
+  },
+  {
+    id: 'provider-006',
+    name: 'privnode',
+    display_name: 'Privnode',
+    description: '',
+    website: 'https://privnode.com',
+    provider_priority: 6,
+    billing_type: 'pay_as_you_go',
+    monthly_used_usd: 0.0,
+    is_active: true,
+    total_endpoints: 0,
+    active_endpoints: 0,
+    total_keys: 0,
+    active_keys: 0,
+    total_models: 6,
+    active_models: 6,
+    avg_health_score: 1.0,
+    unhealthy_endpoints: 0,
+    api_formats: [],
+    endpoint_health_details: [],
+    created_at: '2024-12-07T22:57:18.069024+08:00',
+    updated_at: new Date().toISOString()
+  },
+  {
+    id: 'provider-007',
+    name: 'undying_api',
+    display_name: 'UndyingAPI',
+    description: '',
+    website: 'https://vip.undyingapi.com',
+    provider_priority: 7,
+    billing_type: 'pay_as_you_go',
+    monthly_used_usd: 6.6,
+    is_active: true,
+    total_endpoints: 1,
+    active_endpoints: 1,
+    total_keys: 1,
+    active_keys: 1,
+    total_models: 1,
+    active_models: 1,
+    avg_health_score: 1.0,
+    unhealthy_endpoints: 0,
+    api_formats: ['GEMINI'],
+    endpoint_health_details: [
+      { api_format: 'GEMINI', health_score: 1.0, is_active: true, active_keys: 1 }
+    ],
+    created_at: '2024-12-07T23:00:42.559105+08:00',
+    updated_at: new Date().toISOString()
+  }
+]
+
+// ========== GlobalModel 数据 ==========
+
+export const MOCK_GLOBAL_MODELS: GlobalModelResponse[] = [
+  {
+    id: 'gm-001',
+    name: 'claude-haiku-4-5-20251001',
+    display_name: 'claude-haiku-4-5',
+    description: 'Anthropic 最快速的 Claude 4 系列模型',
+    is_active: true,
+    default_tiered_pricing: {
+      tiers: [{ up_to: null, input_price_per_1m: 1.00, output_price_per_1m: 5.00, cache_creation_price_per_1m: 1.25, cache_read_price_per_1m: 0.1 }]
+    },
+    default_supports_vision: true,
+    default_supports_function_calling: true,
+    default_supports_streaming: true,
+    default_supports_extended_thinking: true,
+    provider_count: 3,
+    alias_count: 2,
+    created_at: '2024-01-01T00:00:00Z'
+  },
+  {
+    id: 'gm-002',
+    name: 'claude-opus-4-5-20251101',
+    display_name: 'claude-opus-4-5',
+    description: 'Anthropic 最强大的模型',
+    is_active: true,
+    default_tiered_pricing: {
+      tiers: [{ up_to: null, input_price_per_1m: 5.00, output_price_per_1m: 25.00, cache_creation_price_per_1m: 6.25, cache_read_price_per_1m: 0.5 }]
+    },
+    default_supports_vision: true,
+    default_supports_function_calling: true,
+    default_supports_streaming: true,
+    default_supports_extended_thinking: true,
+    provider_count: 2,
+    alias_count: 1,
+    created_at: '2024-01-01T00:00:00Z'
+  },
+  {
+    id: 'gm-003',
+    name: 'claude-sonnet-4-5-20250929',
+    display_name: 'claude-sonnet-4-5',
+    description: 'Anthropic 平衡型模型，支持 1h 缓存和 CLI 1M 上下文',
+    is_active: true,
+    default_tiered_pricing: {
+      tiers: [
+        {
+          "up_to": 200000,
+          "input_price_per_1m": 3,
+          "output_price_per_1m": 15,
+          "cache_creation_price_per_1m": 3.75,
+          "cache_read_price_per_1m": 0.3,
+          "cache_ttl_pricing": [
+            {
+              "ttl_minutes": 60,
+              "cache_creation_price_per_1m": 6
+            }
+          ]
+        },
+        {
+          "up_to": null,
+          "input_price_per_1m": 6,
+          "output_price_per_1m": 22.5,
+          "cache_creation_price_per_1m": 7.5,
+          "cache_read_price_per_1m": 0.6,
+          "cache_ttl_pricing": [
+            {
+              "ttl_minutes": 60,
+              "cache_creation_price_per_1m": 12
+            }
+          ]
+        }
+      ]
+    },
+    default_supports_vision: true,
+    default_supports_function_calling: true,
+    default_supports_streaming: true,
+    default_supports_extended_thinking: true,
+    supported_capabilities: ['cache_1h', 'cli_1m'],
+    provider_count: 3,
+    alias_count: 2,
+    created_at: '2024-01-01T00:00:00Z'
+  },
+  {
+    id: 'gm-004',
+    name: 'gemini-3-pro-image-preview',
+    display_name: 'gemini-3-pro-image-preview',
+    description: 'Google Gemini 3 Pro 图像生成预览版',
+    is_active: true,
+    default_price_per_request: 0.300,
+    default_tiered_pricing: {
+      tiers: []
+    },
+    default_supports_vision: true,
+    default_supports_function_calling: false,
+    default_supports_streaming: true,
+    default_supports_image_generation: true,
+    provider_count: 1,
+    alias_count: 0,
+    created_at: '2024-01-01T00:00:00Z'
+  },
+  {
+    id: 'gm-005',
+    name: 'gemini-3-pro-preview',
+    display_name: 'gemini-3-pro-preview',
+    description: 'Google Gemini 3 Pro 预览版',
+    is_active: true,
+    default_tiered_pricing: {
+      tiers: [{ up_to: null, input_price_per_1m: 2.00, output_price_per_1m: 12.00 }]
+    },
+    default_supports_vision: true,
+    default_supports_function_calling: true,
+    default_supports_streaming: true,
+    default_supports_extended_thinking: true,
+    provider_count: 1,
+    alias_count: 0,
+    created_at: '2024-01-01T00:00:00Z'
+  },
+  {
+    id: 'gm-006',
+    name: 'gpt-5.1',
+    display_name: 'gpt-5.1',
+    description: 'OpenAI GPT-5.1 模型',
+    is_active: true,
+    default_tiered_pricing: {
+      tiers: [{ up_to: null, input_price_per_1m: 1.25, output_price_per_1m: 10.00 }]
+    },
+    default_supports_vision: true,
+    default_supports_function_calling: true,
+    default_supports_streaming: true,
+    default_supports_extended_thinking: true,
+    provider_count: 2,
+    alias_count: 1,
+    created_at: '2024-01-01T00:00:00Z'
+  },
+  {
+    id: 'gm-007',
+    name: 'gpt-5.1-codex',
+    display_name: 'gpt-5.1-codex',
+    description: 'OpenAI GPT-5.1 Codex 代码专用模型',
+    is_active: true,
+    default_tiered_pricing: {
+      tiers: [{ up_to: null, input_price_per_1m: 1.25, output_price_per_1m: 10.00 }]
+    },
+    default_supports_vision: true,
+    default_supports_function_calling: true,
+    default_supports_streaming: true,
+    default_supports_extended_thinking: true,
+    provider_count: 2,
+    alias_count: 0,
+    created_at: '2024-01-01T00:00:00Z'
+  },
+  {
+    id: 'gm-008',
+    name: 'gpt-5.1-codex-max',
+    display_name: 'gpt-5.1-codex-max',
+    description: 'OpenAI GPT-5.1 Codex Max 代码专用增强版',
+    is_active: true,
+    default_tiered_pricing: {
+      tiers: [{ up_to: null, input_price_per_1m: 1.25, output_price_per_1m: 10.00 }]
+    },
+    default_supports_vision: true,
+    default_supports_function_calling: true,
+    default_supports_streaming: true,
+    default_supports_extended_thinking: true,
+    provider_count: 2,
+    alias_count: 0,
+    created_at: '2024-01-01T00:00:00Z'
+  },
+  {
+    id: 'gm-009',
+    name: 'gpt-5.1-codex-mini',
+    display_name: 'gpt-5.1-codex-mini',
+    description: 'OpenAI GPT-5.1 Codex Mini 轻量代码模型',
+    is_active: true,
+    default_tiered_pricing: {
+      tiers: [{ up_to: null, input_price_per_1m: 1.25, output_price_per_1m: 10.00 }]
+    },
+    default_supports_vision: true,
+    default_supports_function_calling: true,
+    default_supports_streaming: true,
+    default_supports_extended_thinking: true,
+    provider_count: 2,
+    alias_count: 0,
+    created_at: '2024-01-01T00:00:00Z'
+  }
+]
+
+// ========== Usage 数据 ==========
+
+export const MOCK_USAGE_RESPONSE: UsageResponse = {
+  total_requests: 1234,
+  total_input_tokens: 1800000,
+  total_output_tokens: 700000,
+  total_tokens: 2500000,
+  total_cost: 45.67,
+  total_actual_cost: 33.33,
+  avg_response_time: 1.23,
+  quota_usd: 100,
+  used_usd: 45.32,
+  summary_by_model: [
+    { model: 'claude-sonnet-4-20250514', requests: 456, input_tokens: 650000, output_tokens: 250000, total_tokens: 900000, total_cost_usd: 18.50, actual_total_cost_usd: 13.50 },
+    { model: 'gpt-4o', requests: 312, input_tokens: 480000, output_tokens: 180000, total_tokens: 660000, total_cost_usd: 12.30, actual_total_cost_usd: 9.20 },
+    { model: 'claude-haiku-3-5-20241022', requests: 289, input_tokens: 420000, output_tokens: 170000, total_tokens: 590000, total_cost_usd: 8.50, actual_total_cost_usd: 6.30 },
+    { model: 'gemini-2.0-flash', requests: 177, input_tokens: 250000, output_tokens: 100000, total_tokens: 350000, total_cost_usd: 6.37, actual_total_cost_usd: 4.33 }
+  ],
+  records: [
+    {
+      id: 'usage-001',
+      provider: 'anthropic',
+      model: 'claude-sonnet-4-20250514',
+      input_tokens: 1500,
+      output_tokens: 800,
+      total_tokens: 2300,
+      cost: 0.0165,
+      response_time_ms: 1234,
+      is_stream: true,
+      created_at: new Date().toISOString(),
+      status_code: 200,
+      input_price_per_1m: 3,
+      output_price_per_1m: 15
+    },
+    {
+      id: 'usage-002',
+      provider: 'openai',
+      model: 'gpt-4o',
+      input_tokens: 2000,
+      output_tokens: 500,
+      total_tokens: 2500,
+      cost: 0.01,
+      response_time_ms: 890,
+      is_stream: false,
+      created_at: new Date(Date.now() - 300000).toISOString(),
+      status_code: 200,
+      input_price_per_1m: 2.5,
+      output_price_per_1m: 10
+    }
+  ]
+}
+
+// ========== 系统配置 ==========
+
+export const MOCK_SYSTEM_CONFIGS = [
+  { key: 'rate_limit_enabled', value: true, description: '是否启用速率限制' },
+  { key: 'default_rate_limit', value: 60, description: '默认速率限制（请求/分钟）' },
+  { key: 'cache_enabled', value: true, description: '是否启用缓存' },
+  { key: 'default_cache_ttl', value: 3600, description: '默认缓存 TTL（秒）' },
+  { key: 'fallback_enabled', value: true, description: '是否启用故障转移' },
+  { key: 'max_fallback_attempts', value: 3, description: '最大故障转移次数' }
+]
+
+// ========== API 格式 ==========
+
+export const MOCK_API_FORMATS = {
+  formats: [
+    { value: 'claude', label: 'Claude API', default_path: '/v1/messages', aliases: [] },
+    { value: 'claude_cli', label: 'Claude CLI', default_path: '/v1/messages', aliases: [] },
+    { value: 'openai', label: 'OpenAI API', default_path: '/v1/chat/completions', aliases: [] },
+    { value: 'openai_cli', label: 'OpenAI Responses API', default_path: '/v1/responses', aliases: [] },
+    { value: 'gemini', label: 'Gemini API', default_path: '/v1beta/models', aliases: [] },
+    { value: 'gemini_cli', label: 'Gemini CLI', default_path: '/v1beta/models', aliases: [] }
+  ]
+}

frontend/src/mocks/index.ts

@@ -0,0 +1,6 @@
+/**
+ * Mock Module Index
+ */
+
+export { handleMockRequest, setMockUserToken, getMockUserToken } from './handler'
+export * from './data'

frontend/src/views/shared/Usage.vue

@@ -171,7 +171,14 @@ const filteredRecords = computed(() => {
       } else if (filterStatus.value === 'completed') {
         records = records.filter(record => record.status === 'completed')
       } else if (filterStatus.value === 'failed') {
-        records = records.filter(record => record.status === 'failed')
+        // 失败请求需要同时考虑新旧两种判断方式：
+        // 1. 新方式：status = "failed"
+        // 2. 旧方式：status_code >= 400 或 error_message 不为空
+        records = records.filter(record =>
+          record.status === 'failed' ||
+          (record.status_code && record.status_code >= 400) ||
+          record.error_message
+        )
       }
     }
 

pyproject.toml

@@ -95,6 +95,8 @@ warn_return_any = true
 warn_unused_configs = true
 disallow_untyped_defs = true
 exclude = "tools/debug"
+# 忽略项目内部模块的 import-untyped 警告
+ignore_missing_imports = true
 
 [tool.pytest.ini_options]
 testpaths = ["tests"]

src/api/admin/usage/routes.py

@@ -530,9 +530,18 @@ class AdminUsageRecordsAdapter(AdminApiAdapter):
                 query = query.filter(
                     (Usage.status_code >= 400) | (Usage.error_message.isnot(None))
                 )
-            elif self.status in ("pending", "streaming", "completed", "failed"):
+            elif self.status in ("pending", "streaming", "completed"):
                 # 新的状态筛选：直接按 status 字段过滤
                 query = query.filter(Usage.status == self.status)
+            elif self.status == "failed":
+                # 失败请求需要同时考虑新旧两种判断方式：
+                # 1. 新方式：status = "failed"
+                # 2. 旧方式：status_code >= 400 或 error_message 不为空
+                query = query.filter(
+                    (Usage.status == "failed") |
+                    (Usage.status_code >= 400) |
+                    (Usage.error_message.isnot(None))
+                )
             elif self.status == "active":
                 # 活跃请求：pending 或 streaming 状态
                 query = query.filter(Usage.status.in_(["pending", "streaming"]))
@@ -651,42 +660,17 @@ class AdminActiveRequestsAdapter(AdminApiAdapter):
         self.ids = ids
 
     async def handle(self, context):  # type: ignore[override]
-        db = context.db
+        from src.services.usage import UsageService
 
+        db = context.db
+        id_list = None
         if self.ids:
-            # 查询指定 ID 的请求状态
             id_list = [id.strip() for id in self.ids.split(",") if id.strip()]
             if not id_list:
                 return {"requests": []}
 
-            records = (
+        requests = UsageService.get_active_requests_status(db=db, ids=id_list)
-                db.query(Usage.id, Usage.status, Usage.input_tokens, Usage.output_tokens, Usage.total_cost_usd, Usage.response_time_ms)
+        return {"requests": requests}
-                .filter(Usage.id.in_(id_list))
-                .all()
-            )
-        else:
-            # 查询所有活跃请求（pending 或 streaming）
-            records = (
-                db.query(Usage.id, Usage.status, Usage.input_tokens, Usage.output_tokens, Usage.total_cost_usd, Usage.response_time_ms)
-                .filter(Usage.status.in_(["pending", "streaming"]))
-                .order_by(Usage.created_at.desc())
-                .limit(50)
-                .all()
-            )
-
-        return {
-            "requests": [
-                {
-                    "id": r.id,
-                    "status": r.status,
-                    "input_tokens": r.input_tokens,
-                    "output_tokens": r.output_tokens,
-                    "cost": float(r.total_cost_usd) if r.total_cost_usd else 0,
-                    "response_time_ms": r.response_time_ms,
-                }
-                for r in records
-            ]
-        }
 
 
 @dataclass

src/api/handlers/base/base_handler.py

@@ -50,7 +50,9 @@ class MessageTelemetry:
     负责记录 Usage/Audit，避免处理器里重复代码。
     """
 
-    def __init__(self, db: Session, user, api_key, request_id: str, client_ip: str):
+    def __init__(
+        self, db: Session, user: Any, api_key: Any, request_id: str, client_ip: str
+    ) -> None:
         self.db = db
         self.user = user
         self.api_key = api_key
@@ -187,7 +189,7 @@ class MessageTelemetry:
         response_body: Optional[Dict[str, Any]] = None,
         # 模型映射信息
         target_model: Optional[str] = None,
-    ):
+    ) -> None:
         """
         记录失败请求
 
@@ -283,15 +285,15 @@ class BaseMessageHandler:
         self,
         *,
         db: Session,
-        user,
+        user: Any,
-        api_key,
+        api_key: Any,
         request_id: str,
         client_ip: str,
         user_agent: str,
         start_time: float,
         allowed_api_formats: Optional[list[str]] = None,
         adapter_detector: Optional[AdapterDetectorType] = None,
-    ):
+    ) -> None:
         self.db = db
         self.user = user
         self.api_key = api_key
@@ -304,7 +306,7 @@ class BaseMessageHandler:
         self.adapter_detector = adapter_detector
 
         redis_client = get_redis_client_sync()
-        self.orchestrator = FallbackOrchestrator(db, redis_client)
+        self.orchestrator = FallbackOrchestrator(db, redis_client)  # type: ignore[arg-type]
         self.telemetry = MessageTelemetry(db, user, api_key, request_id, client_ip)
 
     def elapsed_ms(self) -> int:
@@ -347,7 +349,8 @@ class BaseMessageHandler:
     def get_api_format(self, provider_type: Optional[str] = None) -> APIFormat:
         """根据 provider_type 解析 API 格式，未知类型默认 OPENAI"""
         if provider_type:
-            return resolve_api_format(provider_type, default=APIFormat.OPENAI)
+            result = resolve_api_format(provider_type, default=APIFormat.OPENAI)
+            return result or APIFormat.OPENAI
         return self.primary_api_format
 
     def build_provider_payload(
@@ -361,3 +364,34 @@ class BaseMessageHandler:
         if mapped_model:
             payload["model"] = mapped_model
         return payload
+
+    def _update_usage_to_streaming(self, request_id: Optional[str] = None) -> None:
+        """更新 Usage 状态为 streaming（流式传输开始时调用）
+
+        使用 asyncio 后台任务执行数据库更新，避免阻塞流式传输
+
+        Args:
+            request_id: 请求 ID，如果不传则使用 self.request_id
+        """
+        import asyncio
+        from src.database.database import get_db
+
+        target_request_id = request_id or self.request_id
+
+        async def _do_update() -> None:
+            try:
+                db_gen = get_db()
+                db = next(db_gen)
+                try:
+                    UsageService.update_usage_status(
+                        db=db,
+                        request_id=target_request_id,
+                        status="streaming",
+                    )
+                finally:
+                    db.close()
+            except Exception as e:
+                logger.warning(f"[{target_request_id}] 更新 Usage 状态为 streaming 失败: {e}")
+
+        # 创建后台任务，不阻塞当前流
+        asyncio.create_task(_do_update())

src/api/handlers/base/chat_handler_base.py

@@ -116,7 +116,7 @@ class ChatHandlerBase(BaseMessageHandler, ABC):
     # ==================== 抽象方法 ====================
 
     @abstractmethod
-    async def _convert_request(self, request):
+    async def _convert_request(self, request: Any) -> Any:
         """
         将请求转换为目标格式
 
@@ -261,7 +261,7 @@ class ChatHandlerBase(BaseMessageHandler, ABC):
         mapping = await mapper.get_mapping(source_model, provider_id)
 
         if mapping and mapping.model:
-            mapped_name = mapping.model.provider_model_name
+            mapped_name = str(mapping.model.provider_model_name)
             logger.debug(f"[Chat] 模型映射: {source_model} -> {mapped_name}")
             return mapped_name
 
@@ -271,10 +271,10 @@ class ChatHandlerBase(BaseMessageHandler, ABC):
 
     async def process_stream(
         self,
-        request,
+        request: Any,
         http_request: Request,
-        original_headers: Dict,
+        original_headers: Dict[str, Any],
-        original_request_body: Dict,
+        original_request_body: Dict[str, Any],
         query_params: Optional[Dict[str, str]] = None,
     ) -> StreamingResponse:
         """处理流式响应"""
@@ -315,7 +315,7 @@ class ChatHandlerBase(BaseMessageHandler, ABC):
             provider: Provider,
             endpoint: ProviderEndpoint,
             key: ProviderAPIKey,
-        ):
+        ) -> AsyncGenerator[bytes, None]:
             return await self._execute_stream_request(
                 ctx,
                 provider,
@@ -401,16 +401,16 @@ class ChatHandlerBase(BaseMessageHandler, ABC):
         ctx["cached_tokens"] = 0
         ctx["cache_creation_tokens"] = 0
 
-        ctx["provider_name"] = provider.name
+        ctx["provider_name"] = str(provider.name)
-        ctx["provider_id"] = provider.id
+        ctx["provider_id"] = str(provider.id)
-        ctx["endpoint_id"] = endpoint.id
+        ctx["endpoint_id"] = str(endpoint.id)
-        ctx["key_id"] = key.id
+        ctx["key_id"] = str(key.id)
-        ctx["provider_api_format"] = endpoint.api_format  # Provider 的响应格式
+        ctx["provider_api_format"] = str(endpoint.api_format) if endpoint.api_format else ""
 
         # 获取模型映射
         mapped_model = await self._get_mapped_model(
             source_model=ctx["model"],
-            provider_id=provider.id,
+            provider_id=str(provider.id),
         )
 
         # 应用模型映射到请求体
@@ -514,14 +514,20 @@ class ChatHandlerBase(BaseMessageHandler, ABC):
         """创建响应流生成器"""
         try:
             sse_parser = SSEEventParser()
+            streaming_status_updated = False
 
             async for line in stream_response.aiter_lines():
+                # 在第一次输出数据前更新状态为 streaming
+                if not streaming_status_updated:
+                    self._update_usage_to_streaming()
+                    streaming_status_updated = True
+
                 normalized_line = line.rstrip("\r")
                 events = sse_parser.feed_line(normalized_line)
 
                 if normalized_line == "":
                     for event in events:
-                        self._handle_sse_event(ctx, event.get("event"), event.get("data", ""))
+                        self._handle_sse_event(ctx, event.get("event"), event.get("data") or "")
                     yield b"\n"
                     continue
 
@@ -530,11 +536,11 @@ class ChatHandlerBase(BaseMessageHandler, ABC):
                 yield (line + "\n").encode("utf-8")
 
                 for event in events:
-                    self._handle_sse_event(ctx, event.get("event"), event.get("data", ""))
+                    self._handle_sse_event(ctx, event.get("event"), event.get("data") or "")
 
             # 处理剩余事件
             for event in sse_parser.flush():
-                self._handle_sse_event(ctx, event.get("event"), event.get("data", ""))
+                self._handle_sse_event(ctx, event.get("event"), event.get("data") or "")
 
         except GeneratorExit:
             raise
@@ -618,7 +624,7 @@ class ChatHandlerBase(BaseMessageHandler, ABC):
                         f"error_type={parsed.error_type}, "
                         f"message={parsed.error_message}")
                     raise EmbeddedErrorException(
-                        provider_name=provider.name,
+                        provider_name=str(provider.name),
                         error_code=(
                             int(parsed.error_type)
                             if parsed.error_type and parsed.error_type.isdigit()
@@ -652,6 +658,10 @@ class ChatHandlerBase(BaseMessageHandler, ABC):
         try:
             sse_parser = SSEEventParser()
 
+            # 在第一次输出数据前更新状态为 streaming
+            if prefetched_lines:
+                self._update_usage_to_streaming()
+
             # 先输出预读的数据
             for line in prefetched_lines:
                 normalized_line = line.rstrip("\r")
@@ -659,7 +669,7 @@ class ChatHandlerBase(BaseMessageHandler, ABC):
 
                 if normalized_line == "":
                     for event in events:
-                        self._handle_sse_event(ctx, event.get("event"), event.get("data", ""))
+                        self._handle_sse_event(ctx, event.get("event"), event.get("data") or "")
                     yield b"\n"
                     continue
 
@@ -667,7 +677,7 @@ class ChatHandlerBase(BaseMessageHandler, ABC):
                 yield (line + "\n").encode("utf-8")
 
                 for event in events:
-                    self._handle_sse_event(ctx, event.get("event"), event.get("data", ""))
+                    self._handle_sse_event(ctx, event.get("event"), event.get("data") or "")
 
             # 继续输出剩余的流数据（使用同一个迭代器）
             async for line in line_iterator:
@@ -676,7 +686,7 @@ class ChatHandlerBase(BaseMessageHandler, ABC):
 
                 if normalized_line == "":
                     for event in events:
-                        self._handle_sse_event(ctx, event.get("event"), event.get("data", ""))
+                        self._handle_sse_event(ctx, event.get("event"), event.get("data") or "")
                     yield b"\n"
                     continue
 
@@ -684,11 +694,11 @@ class ChatHandlerBase(BaseMessageHandler, ABC):
                 yield (line + "\n").encode("utf-8")
 
                 for event in events:
-                    self._handle_sse_event(ctx, event.get("event"), event.get("data", ""))
+                    self._handle_sse_event(ctx, event.get("event"), event.get("data") or "")
 
             # 处理剩余事件
             for event in sse_parser.flush():
-                self._handle_sse_event(ctx, event.get("event"), event.get("data", ""))
+                self._handle_sse_event(ctx, event.get("event"), event.get("data") or "")
 
         except GeneratorExit:
             raise
@@ -853,8 +863,8 @@ class ChatHandlerBase(BaseMessageHandler, ABC):
 
                 # 根据状态码决定记录成功还是失败
                 # 499 = 客户端断开连接，503 = 服务不可用（如流中断）
-                status_code = ctx.get("status_code")
+                status_code: int = ctx.get("status_code") or 200
-                if status_code and status_code >= 400:
+                if status_code >= 400:
                     # 记录失败的 Usage，但使用已收到的预估 token 信息（来自 message_start）
                     # 这样即使请求中断，也能记录预估成本
                     await bg_telemetry.record_failure(
@@ -955,6 +965,8 @@ class ChatHandlerBase(BaseMessageHandler, ABC):
                 error_message=f"记录统计信息失败: {str(e)[:200]}",
             )
 
+    # _update_usage_to_streaming 方法已移至基类 BaseMessageHandler
+
     async def _update_usage_status_on_error(
         self,
         response_time_ms: int,
@@ -991,11 +1003,11 @@ class ChatHandlerBase(BaseMessageHandler, ABC):
 
             usage = db.query(Usage).filter(Usage.request_id == self.request_id).first()
             if usage:
-                usage.status = status
+                setattr(usage, "status", status)
-                usage.status_code = status_code
+                setattr(usage, "status_code", status_code)
-                usage.response_time_ms = response_time_ms
+                setattr(usage, "response_time_ms", response_time_ms)
                 if error_message:
-                    usage.error_message = error_message
+                    setattr(usage, "error_message", error_message)
                 db.commit()
                 logger.debug(f"[{self.request_id}] Usage 状态已更新: {status}")
         except Exception as e:
@@ -1040,10 +1052,10 @@ class ChatHandlerBase(BaseMessageHandler, ABC):
 
     async def process_sync(
         self,
-        request,
+        request: Any,
         http_request: Request,
-        original_headers: Dict,
+        original_headers: Dict[str, Any],
-        original_request_body: Dict,
+        original_request_body: Dict[str, Any],
         query_params: Optional[Dict[str, str]] = None,
     ) -> JSONResponse:
         """处理非流式响应"""
@@ -1055,31 +1067,31 @@ class ChatHandlerBase(BaseMessageHandler, ABC):
         api_format = self.allowed_api_formats[0]
 
         # 用于跟踪的变量
-        provider_name = None
+        provider_name: Optional[str] = None
-        response_json = None
+        response_json: Optional[Dict[str, Any]] = None
         status_code = 200
-        response_headers = {}
+        response_headers: Dict[str, str] = {}
-        provider_request_headers = {}
+        provider_request_headers: Dict[str, str] = {}
-        provider_request_body = None
+        provider_request_body: Optional[Dict[str, Any]] = None
-        provider_id = None  # Provider ID（用于失败记录）
+        provider_id: Optional[str] = None  # Provider ID（用于失败记录）
-        endpoint_id = None  # Endpoint ID（用于失败记录）
+        endpoint_id: Optional[str] = None  # Endpoint ID（用于失败记录）
-        key_id = None  # Key ID（用于失败记录）
+        key_id: Optional[str] = None  # Key ID（用于失败记录）
-        mapped_model_result = None  # 映射后的目标模型名（用于 Usage 记录）
+        mapped_model_result: Optional[str] = None  # 映射后的目标模型名（用于 Usage 记录）
 
         async def sync_request_func(
             provider: Provider,
             endpoint: ProviderEndpoint,
             key: ProviderAPIKey,
-        ):
+        ) -> Dict[str, Any]:
             nonlocal provider_name, response_json, status_code, response_headers
             nonlocal provider_request_headers, provider_request_body, mapped_model_result
 
-            provider_name = provider.name
+            provider_name = str(provider.name)
 
             # 获取模型映射
             mapped_model = await self._get_mapped_model(
                 source_model=model,
-                provider_id=provider.id,
+                provider_id=str(provider.id),
             )
 
             # 应用模型映射
@@ -1131,7 +1143,7 @@ class ChatHandlerBase(BaseMessageHandler, ABC):
                 elif resp.status_code == 429:
                     raise ProviderRateLimitException(
                         f"提供商速率限制: {provider.name}",
-                        provider_name=provider.name,
+                        provider_name=str(provider.name),
                         response_headers=response_headers,
                     )
                 elif resp.status_code >= 500:
@@ -1142,7 +1154,7 @@ class ChatHandlerBase(BaseMessageHandler, ABC):
                     )
 
                 response_json = resp.json()
-                return response_json
+                return response_json if isinstance(response_json, dict) else {}
 
         try:
             # 解析能力需求
@@ -1170,6 +1182,10 @@ class ChatHandlerBase(BaseMessageHandler, ABC):
             provider_name = actual_provider_name
             response_time_ms = self.elapsed_ms()
 
+            # 确保 response_json 不为 None
+            if response_json is None:
+                response_json = {}
+
             # 规范化响应
             response_json = self._normalize_response(response_json)
 

src/api/handlers/base/cli_handler_base.py

@@ -207,7 +207,7 @@ class CliMessageHandlerBase(BaseMessageHandler):
         logger.debug(f"[CLI] _get_mapped_model: source={source_model}, provider={provider_id[:8]}..., mapping={mapping}")
 
         if mapping and mapping.model:
-            mapped_name = mapping.model.provider_model_name
+            mapped_name = str(mapping.model.provider_model_name)
             logger.debug(f"[CLI] 模型映射: {source_model} -> {mapped_name} (provider={provider_id[:8]}...)")
             return mapped_name
 
@@ -351,7 +351,7 @@ class CliMessageHandlerBase(BaseMessageHandler):
             provider: Provider,
             endpoint: ProviderEndpoint,
             key: ProviderAPIKey,
-        ):
+        ) -> AsyncGenerator[bytes, None]:
             return await self._execute_stream_request(
                 ctx,
                 provider,
@@ -451,19 +451,19 @@ class CliMessageHandlerBase(BaseMessageHandler):
         ctx.response_metadata = {}  # 重置 Provider 响应元数据
 
         # 记录 Provider 信息
-        ctx.provider_name = provider.name
+        ctx.provider_name = str(provider.name)
-        ctx.provider_id = provider.id
+        ctx.provider_id = str(provider.id)
-        ctx.endpoint_id = endpoint.id
+        ctx.endpoint_id = str(endpoint.id)
-        ctx.key_id = key.id
+        ctx.key_id = str(key.id)
 
         # 记录格式转换信息
-        ctx.provider_api_format = endpoint.api_format if endpoint.api_format else ""
+        ctx.provider_api_format = str(endpoint.api_format) if endpoint.api_format else ""
         ctx.client_api_format = ctx.api_format  # 已在 process_stream 中设置
 
         # 获取模型映射（别名/映射 → 实际模型名）
         mapped_model = await self._get_mapped_model(
             source_model=ctx.model,
-            provider_id=provider.id,
+            provider_id=str(provider.id),
         )
 
         # 应用模型映射到请求体（子类可覆盖此方法处理不同格式）
@@ -575,11 +575,17 @@ class CliMessageHandlerBase(BaseMessageHandler):
         try:
             sse_parser = SSEEventParser()
             last_data_time = time.time()
+            streaming_status_updated = False
 
             # 检查是否需要格式转换
             needs_conversion = self._needs_format_conversion(ctx)
 
             async for line in stream_response.aiter_lines():
+                # 在第一次输出数据前更新状态为 streaming
+                if not streaming_status_updated:
+                    self._update_usage_to_streaming(ctx.request_id)
+                    streaming_status_updated = True
+
                 normalized_line = line.rstrip("\r")
                 events = sse_parser.feed_line(normalized_line)
 
@@ -588,7 +594,7 @@ class CliMessageHandlerBase(BaseMessageHandler):
                         self._handle_sse_event(
                             ctx,
                             event.get("event"),
-                            event.get("data", ""),
+                            event.get("data") or "",
                         )
                     yield b"\n"
                     continue
@@ -622,7 +628,7 @@ class CliMessageHandlerBase(BaseMessageHandler):
                     self._handle_sse_event(
                         ctx,
                         event.get("event"),
-                        event.get("data", ""),
+                        event.get("data") or "",
                     )
 
                 if ctx.data_count > 0:
@@ -633,7 +639,7 @@ class CliMessageHandlerBase(BaseMessageHandler):
                 self._handle_sse_event(
                     ctx,
                     event.get("event"),
-                    event.get("data", ""),
+                    event.get("data") or "",
                 )
 
             # 检查是否收到数据
@@ -781,7 +787,7 @@ class CliMessageHandlerBase(BaseMessageHandler):
                         f"error_type={parsed.error_type}, "
                         f"message={parsed.error_message}")
                     raise EmbeddedErrorException(
-                        provider_name=provider.name,
+                        provider_name=str(provider.name),
                         error_code=(
                             int(parsed.error_type)
                             if parsed.error_type and parsed.error_type.isdigit()
@@ -819,6 +825,10 @@ class CliMessageHandlerBase(BaseMessageHandler):
             # 检查是否需要格式转换
             needs_conversion = self._needs_format_conversion(ctx)
 
+            # 在第一次输出数据前更新状态为 streaming
+            if prefetched_lines:
+                self._update_usage_to_streaming(ctx.request_id)
+
             # 先处理预读的数据
             for line in prefetched_lines:
                 normalized_line = line.rstrip("\r")
@@ -829,7 +839,7 @@ class CliMessageHandlerBase(BaseMessageHandler):
                         self._handle_sse_event(
                             ctx,
                             event.get("event"),
-                            event.get("data", ""),
+                            event.get("data") or "",
                         )
                     yield b"\n"
                     continue
@@ -848,7 +858,7 @@ class CliMessageHandlerBase(BaseMessageHandler):
                     self._handle_sse_event(
                         ctx,
                         event.get("event"),
-                        event.get("data", ""),
+                        event.get("data") or "",
                     )
 
                 if ctx.data_count > 0:
@@ -864,7 +874,7 @@ class CliMessageHandlerBase(BaseMessageHandler):
                         self._handle_sse_event(
                             ctx,
                             event.get("event"),
-                            event.get("data", ""),
+                            event.get("data") or "",
                         )
                     yield b"\n"
                     continue
@@ -898,7 +908,7 @@ class CliMessageHandlerBase(BaseMessageHandler):
                     self._handle_sse_event(
                         ctx,
                         event.get("event"),
-                        event.get("data", ""),
+                        event.get("data") or "",
                     )
 
                 if ctx.data_count > 0:
@@ -910,7 +920,7 @@ class CliMessageHandlerBase(BaseMessageHandler):
                 self._handle_sse_event(
                     ctx,
                     event.get("event"),
-                    event.get("data", ""),
+                    event.get("data") or "",
                 )
 
             # 检查是否收到数据
@@ -1270,6 +1280,8 @@ class CliMessageHandlerBase(BaseMessageHandler):
             target_model=ctx.mapped_model,
         )
 
+    # _update_usage_to_streaming 方法已移至基类 BaseMessageHandler
+
     async def process_sync(
         self,
         original_request_body: Dict[str, Any],
@@ -1309,15 +1321,15 @@ class CliMessageHandlerBase(BaseMessageHandler):
             provider: Provider,
             endpoint: ProviderEndpoint,
             key: ProviderAPIKey,
-        ):
+        ) -> Dict[str, Any]:
             nonlocal provider_name, response_json, status_code, response_headers, provider_api_format, provider_request_headers, provider_request_body, mapped_model_result, response_metadata_result
-            provider_name = provider.name
+            provider_name = str(provider.name)
-            provider_api_format = endpoint.api_format if endpoint.api_format else ""
+            provider_api_format = str(endpoint.api_format) if endpoint.api_format else ""
 
             # 获取模型映射（别名/映射 → 实际模型名）
             mapped_model = await self._get_mapped_model(
                 source_model=model,
-                provider_id=provider.id,
+                provider_id=str(provider.id),
             )
 
             # 应用模型映射到请求体（子类可覆盖此方法处理不同格式）
@@ -1373,7 +1385,7 @@ class CliMessageHandlerBase(BaseMessageHandler):
                 elif resp.status_code == 429:
                     raise ProviderRateLimitException(
                         f"提供商速率限制: {provider.name}",
-                        provider_name=provider.name,
+                        provider_name=str(provider.name),
                         response_headers=response_headers,
                         retry_after=int(resp.headers.get("retry-after", 0)) or None,
                     )
@@ -1409,7 +1421,7 @@ class CliMessageHandlerBase(BaseMessageHandler):
                 # 提取 Provider 响应元数据（子类可覆盖）
                 response_metadata_result = self._extract_response_metadata(response_json)
 
-                return response_json
+                return response_json if isinstance(response_json, dict) else {}
 
         try:
             # 解析能力需求
@@ -1437,6 +1449,10 @@ class CliMessageHandlerBase(BaseMessageHandler):
             provider_name = actual_provider_name
             response_time_ms = int((time.time() - sync_start_time) * 1000)
 
+            # 确保 response_json 不为 None
+            if response_json is None:
+                response_json = {}
+
             # 检查是否需要格式转换
             if (
                 provider_api_format

src/api/user_me/routes.py

@@ -662,60 +662,18 @@ class GetActiveRequestsAdapter(AuthenticatedApiAdapter):
     ids: Optional[str] = None
 
     async def handle(self, context):  # type: ignore[override]
+        from src.services.usage import UsageService
+
         db = context.db
         user = context.user
-
+        id_list = None
         if self.ids:
-            # 查询指定 ID 的请求状态（只能查询自己的）
             id_list = [id.strip() for id in self.ids.split(",") if id.strip()]
             if not id_list:
                 return {"requests": []}
 
-            records = (
+        requests = UsageService.get_active_requests_status(db=db, ids=id_list, user_id=user.id)
-                db.query(
+        return {"requests": requests}
-                    Usage.id,
-                    Usage.status,
-                    Usage.input_tokens,
-                    Usage.output_tokens,
-                    Usage.total_cost_usd,
-                    Usage.response_time_ms,
-                )
-                .filter(Usage.id.in_(id_list), Usage.user_id == user.id)
-                .all()
-            )
-        else:
-            # 查询所有活跃请求（pending 或 streaming）
-            records = (
-                db.query(
-                    Usage.id,
-                    Usage.status,
-                    Usage.input_tokens,
-                    Usage.output_tokens,
-                    Usage.total_cost_usd,
-                    Usage.response_time_ms,
-                )
-                .filter(
-                    Usage.user_id == user.id,
-                    Usage.status.in_(["pending", "streaming"]),
-                )
-                .order_by(Usage.created_at.desc())
-                .limit(50)
-                .all()
-            )
-
-        return {
-            "requests": [
-                {
-                    "id": r.id,
-                    "status": r.status,
-                    "input_tokens": r.input_tokens,
-                    "output_tokens": r.output_tokens,
-                    "cost": float(r.total_cost_usd) if r.total_cost_usd else 0,
-                    "response_time_ms": r.response_time_ms,
-                }
-                for r in records
-            ]
-        }
 
 
 class ListAvailableProvidersAdapter(AuthenticatedApiAdapter):

src/services/orchestration/error_classifier.py

@@ -76,6 +76,7 @@ class ErrorClassifier:
         "content_policy_violation",  # 内容违规
         "invalid_api_key",  # 无效的 API Key（不同于认证失败）
         "context_length_exceeded",  # 上下文长度超限
+        "content_length_limit",  # 请求内容长度超限 (Claude API)
         "max_tokens",  # token 数超限
         "invalid_prompt",  # 无效的提示词
         "content too long",  # 内容过长

src/services/usage/service.py

@@ -1304,3 +1304,93 @@ class UsageService:
             )
             .count()
         )
+
+    @classmethod
+    def get_active_requests_status(
+        cls,
+        db: Session,
+        ids: Optional[List[str]] = None,
+        user_id: Optional[str] = None,
+        default_timeout_seconds: int = 300,
+    ) -> List[Dict[str, Any]]:
+        """
+        获取活跃请求状态（用于前端轮询），并自动清理超时的 pending 请求
+
+        与 get_active_requests 不同，此方法：
+        1. 返回轻量级的状态字典而非完整 Usage 对象
+        2. 自动检测并清理超时的 pending 请求
+        3. 支持按 ID 列表查询特定请求
+
+        Args:
+            db: 数据库会话
+            ids: 指定要查询的请求 ID 列表（可选）
+            user_id: 限制只查询该用户的请求（可选，用于普通用户接口）
+            default_timeout_seconds: 默认超时时间（秒），当端点未配置时使用
+
+        Returns:
+            请求状态列表
+        """
+        from src.models.database import ProviderEndpoint
+
+        now = datetime.now(timezone.utc)
+
+        # 构建基础查询，包含端点的 timeout 配置
+        query = db.query(
+            Usage.id,
+            Usage.status,
+            Usage.input_tokens,
+            Usage.output_tokens,
+            Usage.total_cost_usd,
+            Usage.response_time_ms,
+            Usage.created_at,
+            Usage.provider_endpoint_id,
+            ProviderEndpoint.timeout.label("endpoint_timeout"),
+        ).outerjoin(ProviderEndpoint, Usage.provider_endpoint_id == ProviderEndpoint.id)
+
+        if ids:
+            query = query.filter(Usage.id.in_(ids))
+            if user_id:
+                query = query.filter(Usage.user_id == user_id)
+        else:
+            # 查询所有活跃请求
+            query = query.filter(Usage.status.in_(["pending", "streaming"]))
+            if user_id:
+                query = query.filter(Usage.user_id == user_id)
+            query = query.order_by(Usage.created_at.desc()).limit(50)
+
+        records = query.all()
+
+        # 检查超时的 pending 请求
+        timeout_ids = []
+        for r in records:
+            if r.status == "pending" and r.created_at:
+                # 使用端点配置的超时时间，若无则使用默认值
+                timeout_seconds = r.endpoint_timeout or default_timeout_seconds
+
+                # 处理时区：如果 created_at 没有时区信息，假定为 UTC
+                created_at = r.created_at
+                if created_at.tzinfo is None:
+                    created_at = created_at.replace(tzinfo=timezone.utc)
+                elapsed = (now - created_at).total_seconds()
+                if elapsed > timeout_seconds:
+                    timeout_ids.append(r.id)
+
+        # 批量更新超时的请求
+        if timeout_ids:
+            db.query(Usage).filter(Usage.id.in_(timeout_ids)).update(
+                {"status": "failed", "error_message": "请求超时（服务器可能已重启）"},
+                synchronize_session=False,
+            )
+            db.commit()
+
+        return [
+            {
+                "id": r.id,
+                "status": "failed" if r.id in timeout_ids else r.status,
+                "input_tokens": r.input_tokens,
+                "output_tokens": r.output_tokens,
+                "cost": float(r.total_cost_usd) if r.total_cost_usd else 0,
+                "response_time_ms": r.response_time_ms,
+            }
+            for r in records
+        ]