feat: 提示词和 AI 参数移至数据库管理

- AiSetting 模型新增 6 个提示词字段 + 3 个阶段温度字段
- main.py 迁移逻辑改为检测 ai_settings 表的新字段并填充默认值
- generate.py 从数据库读取生成提示词，去掉硬编码
- optimizer 从 API 读取各阶段提示词和温度，删除 prompts.py
- crawler 从 API 读取提取/改写提示词和温度，删除 prompts.py
- settings/active 端点去掉 token 认证（供爬虫/优化器使用）
- 后台设置页新增提示词编辑区和温度调节控件
- 新增 _ensure_default_settings 自动创建默认配置

.claude/settings.local.json

@@ -6,7 +6,15 @@
       "Skill(subagent-driven-development)",
       "Bash(git add *)",
       "Bash(git commit *)",
-      "Bash(python -c \"from app.models import Joke, JokeType, JokeCrowd, AdminUser; print\\('All models imported successfully'\\)\")"
+      "Bash(python -c \"from app.models import Joke, JokeType, JokeCrowd, AdminUser; print\\('All models imported successfully'\\)\")",
+      "Bash(curl -s -o /dev/null -w \"%{http_code}\" http://localhost:8001/health)",
+      "Bash(curl -s http://localhost:8001/api/auth/login -H \"Content-Type: application/json\" -d \"{\\\\\"username\\\\\":\\\\\"admin\\\\\",\\\\\"password\\\\\":\\\\\"admin123\\\\\"}\")",
+      "Bash(python -c \"import sys,json;d=json.load\\(sys.stdin\\);print\\(f'Jokes: {d[\\\\\"total\\\\\"]} total'\\)\")",
+      "Bash(npm run *)",
+      "Bash(curl -s -o /dev/null -w \"%{http_code}\" http://localhost:3000)",
+      "Bash(python -m uvicorn main:app --reload --port 8001)",
+      "Bash(python -c \"from app.schemas.joke import GenerateRequest, GenerateResponse; print\\('OK'\\)\")",
+      "Bash(python -c \"from main import app; print\\('API imports OK'\\)\")"
     ]
   }
 }

CLAUDE.md

@@ -0,0 +1,175 @@
+# CLAUDE.md
+
+本文件为 Claude Code (claude.ai/code) 在此仓库中工作时提供指导。
+
+## 启动命令
+
+### API 后端 (port 8001)
+```bash
+cd api && python -m uvicorn main:app --reload --port 8001
+```
+
+### Web 前台 (port 3000)
+```bash
+cd web && npm run dev      # 开发模式
+cd web && npm run build    # 生产构建
+```
+
+### Admin 后台管理 (port 3001)
+```bash
+cd admin && npm run dev     # 开发模式
+cd admin && npm run build   # 生产构建
+```
+
+### 数据库初始化
+```bash
+cd api && python init_db.py
+```
+
+### 爬虫 (AI 自动采集)
+```bash
+cd crawler && python main.py                     # 持续采集（无头浏览器）
+cd crawler && python main.py --no-headless       # 显示浏览器窗口（调试用）
+```
+
+### 站点发现（独立工具）
+```bash
+python crawler/site_finder.py                    # 搜索笑话站点，保存到 joke_sites.json
+python crawler/site_crawler.py --site https://...  # 采集指定站点
+python crawler/site_crawler.py --file joke_sites.json  # 批量采集已发现站点
+```
+
+### 笑话优化（质量检测 / AI 润色 / 评价分类）
+```bash
+python optimizer/main.py                         # 处理所有笑话
+python optimizer/main.py --limit 10              # 只处理前 10 条（调试用）
+python optimizer/main.py --status rejected       # 只处理已拒绝的
+python optimizer/main.py --id 1,2,3             # 指定 ID 处理
+```
+
+### 启动顺序
+1. API 后端 (port 8001) — 自动建表，必须先启动
+2. Web 前台 (port 3000) — 依赖 API
+3. Admin 后台 (port 3001) — 依赖 API
+4. 爬虫 / 优化器 (可选) — 连接 API 的 8001 端口
+
+## 开发备注
+
+- **无测试**：本项目当前没有自动化测试
+- **数据库迁移**：新增字段需手动 `ALTER TABLE` 或重建，不支持自动迁移
+- **JWT Secret**：开发环境使用默认值，生产环境需设置 `JWT_SECRET_KEY` 环境变量
+
+## 项目架构
+
+### 目录结构
+```
+joke/
+├── api/                    # FastAPI 后端 (Python, port 8001)
+│   └── app/
+│       ├── models/         # SQLAlchemy 模型（自动建表）
+│       ├── routers/        # API 路由
+│       └── schemas/        # Pydantic 数据模型
+├── web/                    # Vue3 前台 (port 3000)
+│   └── src/
+│       ├── components/     # 通用组件（三栏布局）
+│       ├── views/          # 页面（首页/分类/详情/搜索）
+│       ├── stores/         # Pinia（joke/category/theme）
+│       └── api/            # Axios 封装
+├── admin/                  # Vue3 管理后台 (port 3001)
+│   └── src/
+│       ├── layout/         # 后台布局（侧边栏+顶部导航）
+│       ├── views/          # 页面（登录/仪表盘/笑话管理/分类管理/设置）
+│       ├── stores/         # Pinia（auth/joke）
+│       └── api/            # Axios 封装
+├── crawler/                # AI 爬虫（crawl4ai + NVIDIA NIM）
+│   ├── main.py             # 持续采集主入口
+│   ├── processor.py        # 流程编排（搜索→深度翻页→AI提取→入库）
+│   ├── crawler_service.py  # 页面抓取（Bing搜索 + crawl4ai）
+│   ├── ai_service.py       # NVIDIA NIM API（OpenAI 兼容）调用
+│   ├── prompts.py          # AI 提示词模板
+│   ├── site_finder.py      # 站点发现工具（搜索 Bing 找笑话站）
+│   └── site_crawler.py     # 单站点批量采集工具
+├── optimizer/              # 笑话优化工具（质量检测→润色→评价分类）
+│   ├── main.py             # CLI 入口
+│   ├── optimizer.py        # 三阶段处理管道
+│   └── prompts.py          # AI 提示词模板
+└── docs/                   # 文档
+```
+
+### 关键约定
+
+- **API 代理**：web 和 admin 的 Vite 配置都将 `/api` 代理到 `http://localhost:8001`，不重写路径
+- **JWT 认证**：登录接口 `/api/auth/login`，token 存 localStorage，请求头 `Authorization: Bearer <token>`
+- **笑话审核流**：`pending`（待审核）→ `approved`（已通过）。公开 API 只返回 `approved` 的笑话
+- **双主题系统**：CSS 变量通过 `:root[data-theme="light"]` / `:root[data-theme="dark"]` 控制，使用 `data-theme` 属性切换
+- **数据库**：SQLite (`api/joke.db`)，SQLAlchemy ORM，API 启动时自动建表（`Base.metadata.create_all`），但**不自动修改已有表结构**——新增字段需手动 `ALTER TABLE` 或重建
+- **爬虫**：使用 crawl4ai 的 `AsyncWebCrawler`（统一 headless/visible 模式），crawl4ai 不可用时回退到 requests
+- **AI 服务**：使用 NVIDIA NIM API（兼容 OpenAI），配置通过后台设置页面管理，存储在 `AiSetting` 表
+
+### 数据模型
+
+| 表名 | 说明 | 关键字段 |
+|------|------|----------|
+| `jokes` | 笑话 | title, content, polished_content, type_id, crowd_id, status, view_count, like_count |
+| `joke_types` | 类型 | name, icon, sort_order |
+| `joke_crowds` | 人群 | name, icon, sort_order |
+| `admin_users` | 管理员 | username, password_hash（bcrypt） |
+| `ai_settings` | AI 配置 | api_base, api_key, model_name, temperature, is_active |
+| `links` | 友链 | name, url, status |
+| `feedback` | 用户反馈 | content, contact, status |
+
+### API 接口一览
+
+- `GET /api/jokes/` — 公开：获取已审核笑话列表（支持 `type_id`/`crowd_id` 筛选，分页）
+- `GET /api/jokes/{id}` — 公开：获取单条笑话（自动增加浏览次数）
+- `GET /api/jokes/random` — 公开：随机获取一条
+- `POST /api/jokes/{id}/like` — 公开：为笑话点赞（增加 like_count）
+- `POST /api/generate` — 公开：AI 生成笑话（keywords + scenarios）
+- `GET /api/categories/types` — 公开：类型列表
+- `GET /api/categories/crowds` — 公开：人群列表
+- `POST /api/auth/login` — 公开：管理员登录，返回 JWT token
+- `GET/POST /api/admin/jokes` — 需认证：列表（支持 `status` 筛选）/ 创建
+- `GET/PUT/DELETE /api/admin/jokes/{id}` — 需认证：单条 CRUD
+- `PUT /api/admin/jokes/batch-approve` — 需认证：批量审核通过
+- `GET /api/admin/stats` — 需认证：统计数据
+- `GET/PUT /api/admin/settings/active` — 需认证：AI 配置管理
+- `GET/POST /api/admin/links` — 需认证：友链管理
+- `GET/POST /api/admin/feedback` — 需认证：反馈管理
+
+**API 文档**：访问 `/docs` 查看交互式 Swagger UI（FastAPI 自动生成）
+
+### 前端三栏布局（web）
+
+- **AppSidebar**：左侧宽频/点击/搞笑段子排行
+- **AppHeader**：顶部导航 + 分类标签 + 搜索框 + 主题切换
+- **AppSubNav**：内容区顶部二级导航（最新/最热/随机）
+- **主内容区**：JokeCard 列表（无限滚动 + 点赞 + 分类/人群标签）
+- **AppRightbar**：右侧类型/人群分类导航
+- **AppFooter**：底部信息
+
+### 爬虫深度采集流程
+
+1. Bing 搜索关键词 → 发现笑话聚合站
+2. 抓取首页 → AI 提取笑话 → 入库
+3. 发现翻页链接（`page_N.html`、`?page=N` 等）→ 逐页抓取 + AI 提取
+4. 发现分类链接（`category-N.html`）→ 逐类深度采集（含分类翻页 `category-N_M.html`）
+5. 站点容错：连续失败 3 次自动跳过
+6. 去重：基于 content MD5 哈希
+
+### 优化器三阶段流程
+
+1. **质量检测** (temperature=0.3) — AI 判断是否有笑点，无则标记 `rejected`
+2. **AI 润色** (temperature=0.8) — 优化语言表达，保存到 `polished_content` 字段
+3. **评价分类** (temperature=0.3) — 分配 type/crowd，评分 1-10，决定 `approved`/`pending`
+
+### 爬虫/优化器通用模式（复用方式）
+
+所有 Python CLI 工具遵循相同模式：
+- 通过 API（非直连数据库）读写数据
+- 复用 `httpx` 进行 REST 调用 + Bearer token 认证
+- AI 调用使用 `openai.OpenAI` 客户端
+- 分类映射：通过 `GET /api/categories/types` 和 `GET /api/categories/crowds` 获取 name→id 映射
+
+---
+
+如果有任何问题或需要帮助，请随时告知！

__joke.json

@@ -0,0 +1 @@
+{"title":"程序员的幽默","content":"程序员去相亲，女方问：你有什么优点？程序员答：我bug少。女方：...","type_id":2,"crowd_id":2,"id":1,"status":"approved","view_count":2,"like_count":0,"created_at":"2026-05-21T14:05:38","updated_at":"2026-05-22T10:40:03","type_name":"谐音梗","crowd_name":"职场"}

__login.json

@@ -0,0 +1 @@
+{"access_token":"eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJzdWIiOiIxIiwidXNlcm5hbWUiOiJhZG1pbiIsImV4cCI6MTc4MDM1NDE1OX0.jqZHtndPbRhPYY7ZrkRQGqLzAMtfbejzkhRCq7fwqGY","token_type":"bearer"}

admin/node_modules/.vite/deps/_metadata.json

@@ -0,0 +1,49 @@
+{
+  "hash": "529e4e6a",
+  "configHash": "47bdc112",
+  "lockfileHash": "81bd2f5f",
+  "browserHash": "e99f2664",
+  "optimized": {
+    "axios": {
+      "src": "../../axios/index.js",
+      "file": "axios.js",
+      "fileHash": "64a84452",
+      "needsInterop": false
+    },
+    "element-plus": {
+      "src": "../../element-plus/es/index.mjs",
+      "file": "element-plus.js",
+      "fileHash": "d8579f02",
+      "needsInterop": false
+    },
+    "pinia": {
+      "src": "../../pinia/dist/pinia.mjs",
+      "file": "pinia.js",
+      "fileHash": "cc8debcd",
+      "needsInterop": false
+    },
+    "vue": {
+      "src": "../../vue/dist/vue.runtime.esm-bundler.js",
+      "file": "vue.js",
+      "fileHash": "7c6ca729",
+      "needsInterop": false
+    },
+    "vue-router": {
+      "src": "../../vue-router/dist/vue-router.mjs",
+      "file": "vue-router.js",
+      "fileHash": "bb6a478b",
+      "needsInterop": false
+    }
+  },
+  "chunks": {
+    "chunk-YFT6OQ5R": {
+      "file": "chunk-YFT6OQ5R.js"
+    },
+    "chunk-UC4PQXLF": {
+      "file": "chunk-UC4PQXLF.js"
+    },
+    "chunk-G3PMV62Z": {
+      "file": "chunk-G3PMV62Z.js"
+    }
+  }
+}

admin/node_modules/.vite/deps/chunk-G3PMV62Z.js

@@ -0,0 +1,36 @@
+var __create = Object.create;
+var __defProp = Object.defineProperty;
+var __getOwnPropDesc = Object.getOwnPropertyDescriptor;
+var __getOwnPropNames = Object.getOwnPropertyNames;
+var __getProtoOf = Object.getPrototypeOf;
+var __hasOwnProp = Object.prototype.hasOwnProperty;
+var __commonJS = (cb, mod) => function __require() {
+  return mod || (0, cb[__getOwnPropNames(cb)[0]])((mod = { exports: {} }).exports, mod), mod.exports;
+};
+var __export = (target, all) => {
+  for (var name in all)
+    __defProp(target, name, { get: all[name], enumerable: true });
+};
+var __copyProps = (to, from, except, desc) => {
+  if (from && typeof from === "object" || typeof from === "function") {
+    for (let key of __getOwnPropNames(from))
+      if (!__hasOwnProp.call(to, key) && key !== except)
+        __defProp(to, key, { get: () => from[key], enumerable: !(desc = __getOwnPropDesc(from, key)) || desc.enumerable });
+  }
+  return to;
+};
+var __toESM = (mod, isNodeMode, target) => (target = mod != null ? __create(__getProtoOf(mod)) : {}, __copyProps(
+  // If the importer is in node compatibility mode or this is not an ESM
+  // file that has been converted to a CommonJS file using a Babel-
+  // compatible transform (i.e. "__esModule" has not been set), then set
+  // "default" to the CommonJS "module.exports" for node compatibility.
+  isNodeMode || !mod || !mod.__esModule ? __defProp(target, "default", { value: mod, enumerable: true }) : target,
+  mod
+));
+
+export {
+  __commonJS,
+  __export,
+  __toESM
+};
+//# sourceMappingURL=chunk-G3PMV62Z.js.map

admin/node_modules/.vite/deps/chunk-G3PMV62Z.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": [],
+  "sourcesContent": [],
+  "mappings": "",
+  "names": []
+}

admin/node_modules/.vite/deps/chunk-YFT6OQ5R.js

@@ -0,0 +1,162 @@
+// node_modules/@vue/devtools-api/lib/esm/env.js
+function getDevtoolsGlobalHook() {
+  return getTarget().__VUE_DEVTOOLS_GLOBAL_HOOK__;
+}
+function getTarget() {
+  return typeof navigator !== "undefined" && typeof window !== "undefined" ? window : typeof globalThis !== "undefined" ? globalThis : {};
+}
+var isProxyAvailable = typeof Proxy === "function";
+
+// node_modules/@vue/devtools-api/lib/esm/const.js
+var HOOK_SETUP = "devtools-plugin:setup";
+var HOOK_PLUGIN_SETTINGS_SET = "plugin:settings:set";
+
+// node_modules/@vue/devtools-api/lib/esm/time.js
+var supported;
+var perf;
+function isPerformanceSupported() {
+  var _a;
+  if (supported !== void 0) {
+    return supported;
+  }
+  if (typeof window !== "undefined" && window.performance) {
+    supported = true;
+    perf = window.performance;
+  } else if (typeof globalThis !== "undefined" && ((_a = globalThis.perf_hooks) === null || _a === void 0 ? void 0 : _a.performance)) {
+    supported = true;
+    perf = globalThis.perf_hooks.performance;
+  } else {
+    supported = false;
+  }
+  return supported;
+}
+function now() {
+  return isPerformanceSupported() ? perf.now() : Date.now();
+}
+
+// node_modules/@vue/devtools-api/lib/esm/proxy.js
+var ApiProxy = class {
+  constructor(plugin, hook) {
+    this.target = null;
+    this.targetQueue = [];
+    this.onQueue = [];
+    this.plugin = plugin;
+    this.hook = hook;
+    const defaultSettings = {};
+    if (plugin.settings) {
+      for (const id in plugin.settings) {
+        const item = plugin.settings[id];
+        defaultSettings[id] = item.defaultValue;
+      }
+    }
+    const localSettingsSaveId = `__vue-devtools-plugin-settings__${plugin.id}`;
+    let currentSettings = Object.assign({}, defaultSettings);
+    try {
+      const raw = localStorage.getItem(localSettingsSaveId);
+      const data = JSON.parse(raw);
+      Object.assign(currentSettings, data);
+    } catch (e) {
+    }
+    this.fallbacks = {
+      getSettings() {
+        return currentSettings;
+      },
+      setSettings(value) {
+        try {
+          localStorage.setItem(localSettingsSaveId, JSON.stringify(value));
+        } catch (e) {
+        }
+        currentSettings = value;
+      },
+      now() {
+        return now();
+      }
+    };
+    if (hook) {
+      hook.on(HOOK_PLUGIN_SETTINGS_SET, (pluginId, value) => {
+        if (pluginId === this.plugin.id) {
+          this.fallbacks.setSettings(value);
+        }
+      });
+    }
+    this.proxiedOn = new Proxy({}, {
+      get: (_target, prop) => {
+        if (this.target) {
+          return this.target.on[prop];
+        } else {
+          return (...args) => {
+            this.onQueue.push({
+              method: prop,
+              args
+            });
+          };
+        }
+      }
+    });
+    this.proxiedTarget = new Proxy({}, {
+      get: (_target, prop) => {
+        if (this.target) {
+          return this.target[prop];
+        } else if (prop === "on") {
+          return this.proxiedOn;
+        } else if (Object.keys(this.fallbacks).includes(prop)) {
+          return (...args) => {
+            this.targetQueue.push({
+              method: prop,
+              args,
+              resolve: () => {
+              }
+            });
+            return this.fallbacks[prop](...args);
+          };
+        } else {
+          return (...args) => {
+            return new Promise((resolve) => {
+              this.targetQueue.push({
+                method: prop,
+                args,
+                resolve
+              });
+            });
+          };
+        }
+      }
+    });
+  }
+  async setRealTarget(target) {
+    this.target = target;
+    for (const item of this.onQueue) {
+      this.target.on[item.method](...item.args);
+    }
+    for (const item of this.targetQueue) {
+      item.resolve(await this.target[item.method](...item.args));
+    }
+  }
+};
+
+// node_modules/@vue/devtools-api/lib/esm/index.js
+function setupDevtoolsPlugin(pluginDescriptor, setupFn) {
+  const descriptor = pluginDescriptor;
+  const target = getTarget();
+  const hook = getDevtoolsGlobalHook();
+  const enableProxy = isProxyAvailable && descriptor.enableEarlyProxy;
+  if (hook && (target.__VUE_DEVTOOLS_PLUGIN_API_AVAILABLE__ || !enableProxy)) {
+    hook.emit(HOOK_SETUP, pluginDescriptor, setupFn);
+  } else {
+    const proxy = enableProxy ? new ApiProxy(descriptor, hook) : null;
+    const list = target.__VUE_DEVTOOLS_PLUGINS__ = target.__VUE_DEVTOOLS_PLUGINS__ || [];
+    list.push({
+      pluginDescriptor: descriptor,
+      setupFn,
+      proxy
+    });
+    if (proxy) {
+      setupFn(proxy.proxiedTarget);
+    }
+  }
+}
+
+export {
+  setupDevtoolsPlugin
+};
+//# sourceMappingURL=chunk-YFT6OQ5R.js.map

admin/node_modules/.vite/deps/package.json

@@ -0,0 +1,3 @@
+{
+  "type": "module"
+}

admin/node_modules/.vite/deps/vue.js

@@ -0,0 +1,348 @@
+import {
+  BaseTransition,
+  BaseTransitionPropsValidators,
+  Comment,
+  DeprecationTypes,
+  EffectScope,
+  ErrorCodes,
+  ErrorTypeStrings,
+  Fragment,
+  KeepAlive,
+  ReactiveEffect,
+  Static,
+  Suspense,
+  Teleport,
+  Text,
+  TrackOpTypes,
+  Transition,
+  TransitionGroup,
+  TriggerOpTypes,
+  VueElement,
+  assertNumber,
+  callWithAsyncErrorHandling,
+  callWithErrorHandling,
+  camelize,
+  capitalize,
+  cloneVNode,
+  compatUtils,
+  compile,
+  computed,
+  createApp,
+  createBaseVNode,
+  createBlock,
+  createCommentVNode,
+  createElementBlock,
+  createHydrationRenderer,
+  createPropsRestProxy,
+  createRenderer,
+  createSSRApp,
+  createSlots,
+  createStaticVNode,
+  createTextVNode,
+  createVNode,
+  customRef,
+  defineAsyncComponent,
+  defineComponent,
+  defineCustomElement,
+  defineEmits,
+  defineExpose,
+  defineModel,
+  defineOptions,
+  defineProps,
+  defineSSRCustomElement,
+  defineSlots,
+  devtools,
+  effect,
+  effectScope,
+  getCurrentInstance,
+  getCurrentScope,
+  getCurrentWatcher,
+  getTransitionRawChildren,
+  guardReactiveProps,
+  h,
+  handleError,
+  hasInjectionContext,
+  hydrate,
+  hydrateOnIdle,
+  hydrateOnInteraction,
+  hydrateOnMediaQuery,
+  hydrateOnVisible,
+  initCustomFormatter,
+  initDirectivesForSSR,
+  inject,
+  isMemoSame,
+  isProxy,
+  isReactive,
+  isReadonly,
+  isRef,
+  isRuntimeOnly,
+  isShallow,
+  isVNode,
+  markRaw,
+  mergeDefaults,
+  mergeModels,
+  mergeProps,
+  nextTick,
+  nodeOps,
+  normalizeClass,
+  normalizeProps,
+  normalizeStyle,
+  onActivated,
+  onBeforeMount,
+  onBeforeUnmount,
+  onBeforeUpdate,
+  onDeactivated,
+  onErrorCaptured,
+  onMounted,
+  onRenderTracked,
+  onRenderTriggered,
+  onScopeDispose,
+  onServerPrefetch,
+  onUnmounted,
+  onUpdated,
+  onWatcherCleanup,
+  openBlock,
+  patchProp,
+  popScopeId,
+  provide,
+  proxyRefs,
+  pushScopeId,
+  queuePostFlushCb,
+  reactive,
+  readonly,
+  ref,
+  registerRuntimeCompiler,
+  render,
+  renderList,
+  renderSlot,
+  resolveComponent,
+  resolveDirective,
+  resolveDynamicComponent,
+  resolveFilter,
+  resolveTransitionHooks,
+  setBlockTracking,
+  setDevtoolsHook,
+  setTransitionHooks,
+  shallowReactive,
+  shallowReadonly,
+  shallowRef,
+  ssrContextKey,
+  ssrUtils,
+  stop,
+  toDisplayString,
+  toHandlerKey,
+  toHandlers,
+  toRaw,
+  toRef,
+  toRefs,
+  toValue,
+  transformVNodeArgs,
+  triggerRef,
+  unref,
+  useAttrs,
+  useCssModule,
+  useCssVars,
+  useHost,
+  useId,
+  useModel,
+  useSSRContext,
+  useShadowRoot,
+  useSlots,
+  useTemplateRef,
+  useTransitionState,
+  vModelCheckbox,
+  vModelDynamic,
+  vModelRadio,
+  vModelSelect,
+  vModelText,
+  vShow,
+  version,
+  warn,
+  watch,
+  watchEffect,
+  watchPostEffect,
+  watchSyncEffect,
+  withAsyncContext,
+  withCtx,
+  withDefaults,
+  withDirectives,
+  withKeys,
+  withMemo,
+  withModifiers,
+  withScopeId
+} from "./chunk-UC4PQXLF.js";
+import "./chunk-G3PMV62Z.js";
+export {
+  BaseTransition,
+  BaseTransitionPropsValidators,
+  Comment,
+  DeprecationTypes,
+  EffectScope,
+  ErrorCodes,
+  ErrorTypeStrings,
+  Fragment,
+  KeepAlive,
+  ReactiveEffect,
+  Static,
+  Suspense,
+  Teleport,
+  Text,
+  TrackOpTypes,
+  Transition,
+  TransitionGroup,
+  TriggerOpTypes,
+  VueElement,
+  assertNumber,
+  callWithAsyncErrorHandling,
+  callWithErrorHandling,
+  camelize,
+  capitalize,
+  cloneVNode,
+  compatUtils,
+  compile,
+  computed,
+  createApp,
+  createBlock,
+  createCommentVNode,
+  createElementBlock,
+  createBaseVNode as createElementVNode,
+  createHydrationRenderer,
+  createPropsRestProxy,
+  createRenderer,
+  createSSRApp,
+  createSlots,
+  createStaticVNode,
+  createTextVNode,
+  createVNode,
+  customRef,
+  defineAsyncComponent,
+  defineComponent,
+  defineCustomElement,
+  defineEmits,
+  defineExpose,
+  defineModel,
+  defineOptions,
+  defineProps,
+  defineSSRCustomElement,
+  defineSlots,
+  devtools,
+  effect,
+  effectScope,
+  getCurrentInstance,
+  getCurrentScope,
+  getCurrentWatcher,
+  getTransitionRawChildren,
+  guardReactiveProps,
+  h,
+  handleError,
+  hasInjectionContext,
+  hydrate,
+  hydrateOnIdle,
+  hydrateOnInteraction,
+  hydrateOnMediaQuery,
+  hydrateOnVisible,
+  initCustomFormatter,
+  initDirectivesForSSR,
+  inject,
+  isMemoSame,
+  isProxy,
+  isReactive,
+  isReadonly,
+  isRef,
+  isRuntimeOnly,
+  isShallow,
+  isVNode,
+  markRaw,
+  mergeDefaults,
+  mergeModels,
+  mergeProps,
+  nextTick,
+  nodeOps,
+  normalizeClass,
+  normalizeProps,
+  normalizeStyle,
+  onActivated,
+  onBeforeMount,
+  onBeforeUnmount,
+  onBeforeUpdate,
+  onDeactivated,
+  onErrorCaptured,
+  onMounted,
+  onRenderTracked,
+  onRenderTriggered,
+  onScopeDispose,
+  onServerPrefetch,
+  onUnmounted,
+  onUpdated,
+  onWatcherCleanup,
+  openBlock,
+  patchProp,
+  popScopeId,
+  provide,
+  proxyRefs,
+  pushScopeId,
+  queuePostFlushCb,
+  reactive,
+  readonly,
+  ref,
+  registerRuntimeCompiler,
+  render,
+  renderList,
+  renderSlot,
+  resolveComponent,
+  resolveDirective,
+  resolveDynamicComponent,
+  resolveFilter,
+  resolveTransitionHooks,
+  setBlockTracking,
+  setDevtoolsHook,
+  setTransitionHooks,
+  shallowReactive,
+  shallowReadonly,
+  shallowRef,
+  ssrContextKey,
+  ssrUtils,
+  stop,
+  toDisplayString,
+  toHandlerKey,
+  toHandlers,
+  toRaw,
+  toRef,
+  toRefs,
+  toValue,
+  transformVNodeArgs,
+  triggerRef,
+  unref,
+  useAttrs,
+  useCssModule,
+  useCssVars,
+  useHost,
+  useId,
+  useModel,
+  useSSRContext,
+  useShadowRoot,
+  useSlots,
+  useTemplateRef,
+  useTransitionState,
+  vModelCheckbox,
+  vModelDynamic,
+  vModelRadio,
+  vModelSelect,
+  vModelText,
+  vShow,
+  version,
+  warn,
+  watch,
+  watchEffect,
+  watchPostEffect,
+  watchSyncEffect,
+  withAsyncContext,
+  withCtx,
+  withDefaults,
+  withDirectives,
+  withKeys,
+  withMemo,
+  withModifiers,
+  withScopeId
+};
+//# sourceMappingURL=vue.js.map

admin/node_modules/.vite/deps/vue.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": [],
+  "sourcesContent": [],
+  "mappings": "",
+  "names": []
+}

admin/src/api/category.js

@@ -1,7 +1,7 @@
 import request from './request'
 
-export const getTypes = () => request.get('/types')
-export const getCrowds = () => request.get('/crowds')
+export const getTypes = () => request.get('/categories/types')
+export const getCrowds = () => request.get('/categories/crowds')
 export const createType = data => request.post('/admin/types', data)
 export const updateType = (id, data) => request.put(`/admin/types/${id}`, data)
 export const deleteType = id => request.delete(`/admin/types/${id}`)

admin/src/api/feedback.js

@@ -0,0 +1,4 @@
+import request from './request'
+
+export const getFeedbacks = () => request.get('/admin/feedbacks')
+export const deleteFeedback = id => request.delete('/admin/feedbacks/' + id)

admin/src/api/link.js

@@ -0,0 +1,7 @@
+import request from './request'
+
+export const getLinks = () => request.get('/admin/links')
+export const getLink = id => request.get('/admin/links/' + id)
+export const createLink = data => request.post('/admin/links', data)
+export const updateLink = (id, data) => request.put('/admin/links/' + id, data)
+export const deleteLink = id => request.delete('/admin/links/' + id)

admin/src/api/request.js

@@ -13,15 +13,15 @@ request.interceptors.request.use(config => {
   return config
 })
 
-request.interceptors.response.use({
-  successHandler: res => res.data,
-  errorHandler: err => {
+request.interceptors.response.use(
+  res => res.data,
+  err => {
     if (err.response?.status === 401) {
       localStorage.removeItem('token')
-      window.location.href = '/login'
+      window.location.href = '/admin/login'
     }
     return Promise.reject(err)
   }
-})
+)
 
 export default request

admin/src/api/setting.js

@@ -0,0 +1,8 @@
+import request from './request'
+
+export const getSettings = () => request.get('/admin/settings')
+export const getActiveSetting = () => request.get('/admin/settings/active')
+export const createSetting = data => request.post('/admin/settings', data)
+export const updateSetting = (id, data) => request.put(`/admin/settings/${id}`, data)
+export const toggleSetting = id => request.put(`/admin/settings/${id}/toggle`)
+export const deleteSetting = id => request.delete(`/admin/settings/${id}`)

admin/src/api/user.js

@@ -0,0 +1,7 @@
+import request from './request'
+
+export const getUsers = () => request.get('/admin/users')
+
+export const createUser = (data) => request.post('/admin/users', data)
+
+export const deleteUser = (id) => request.delete(`/admin/users/${id}`)

admin/src/layout/index.vue

@@ -1,26 +1,40 @@
 <template>
   <el-container style="min-height: 100vh">
-    <el-aside width="200px" style="background: #304156">
+    <el-aside width="220px" style="background: #304156">
       <div class="logo">笑话管理后台</div>
       <el-menu
         :default-active="$route.path"
+        :unique-opened="true"
         background-color="#304156"
         text-color="#bfcbd9"
         active-text-color="#409EFF"
         router
       >
-        <el-menu-item index="/dashboard">
-          <span>仪表盘</span>
-        </el-menu-item>
-        <el-menu-item index="/jokes">
-          <span>笑话管理</span>
-        </el-menu-item>
-        <el-menu-item index="/category/type">
-          <span>类型管理</span>
-        </el-menu-item>
-        <el-menu-item index="/category/crowd">
-          <span>人群管理</span>
-        </el-menu-item>
+        <el-sub-menu index="/dashboard">
+          <template #title>
+            <span>首页</span>
+          </template>
+          <el-menu-item index="/dashboard">仪表盘</el-menu-item>
+        </el-sub-menu>
+
+        <el-sub-menu index="/content">
+          <template #title>
+            <span>内容管理</span>
+          </template>
+          <el-menu-item index="/jokes">笑话管理</el-menu-item>
+          <el-menu-item index="/category/type">类型管理</el-menu-item>
+          <el-menu-item index="/category/crowd">人群管理</el-menu-item>
+        </el-sub-menu>
+
+        <el-sub-menu index="/system">
+          <template #title>
+            <span>系统管理</span>
+          </template>
+          <el-menu-item index="/settings">AI 配置</el-menu-item>
+          <el-menu-item index="/links">友链管理</el-menu-item>
+          <el-menu-item index="/feedbacks">反馈管理</el-menu-item>
+          <el-menu-item index="/users">用户管理</el-menu-item>
+        </el-sub-menu>
       </el-menu>
     </el-aside>
     <el-container>
@@ -56,8 +70,26 @@ const handleLogout = () => {
   font-size: 18px;
   font-weight: bold;
   background: #2b3a4a;
+  letter-spacing: 2px;
 }
 .el-aside {
   background: #304156;
 }
+/* 多级菜单样式 */
+:deep(.el-sub-menu__title) {
+  font-size: 14px;
+  font-weight: 600;
+}
+:deep(.el-menu-item) {
+  font-size: 13px;
+  padding-left: 54px !important;
+}
+:deep(.el-menu-item.is-active) {
+  background-color: #263445 !important;
+}
+:deep(.el-sub-menu .el-menu-item) {
+  min-width: 0;
+  height: 44px;
+  line-height: 44px;
+}
 </style>

admin/src/router/index.js

@@ -16,12 +16,17 @@ const routes = [
       { path: 'jokes/edit/:id?', name: 'JokeEdit', component: () => import('@/views/joke/Edit.vue') },
       { path: 'category/type', name: 'CategoryType', component: () => import('@/views/category/Type.vue') },
       { path: 'category/crowd', name: 'CategoryCrowd', component: () => import('@/views/category/Crowd.vue') },
+      { path: 'settings', name: 'Settings', component: () => import('@/views/setting/Index.vue') },
+      { path: 'links', name: 'LinkList', component: () => import('@/views/links/List.vue') },
+      { path: 'links/edit/:id?', name: 'LinkEdit', component: () => import('@/views/links/Edit.vue') },
+      { path: 'feedbacks', name: 'FeedbackList', component: () => import('@/views/feedback/List.vue') },
+      { path: 'users', name: 'UserList', component: () => import('@/views/user/List.vue') },
     ]
   }
 ]
 
 const router = createRouter({
-  history: createWebHistory(),
+  history: createWebHistory('/admin'),
   routes
 })
 

admin/src/stores/joke.js

@@ -11,7 +11,10 @@ export const useJokeStore = defineStore('joke', {
     async fetchJokes(params) {
       this.loading = true
       try {
-        const res = await getJokes(params)
+        const cleanParams = Object.fromEntries(
+          Object.entries(params).filter(([_, v]) => v != null && v !== '')
+        )
+        const res = await getJokes(cleanParams)
         this.jokes = res.items
         this.total = res.total
       } finally {

admin/src/views/category/Crowd.vue

@@ -4,7 +4,23 @@
       <h2>人群管理</h2>
       <el-button type="primary" @click="dialogVisible = true; form = {}">添加</el-button>
     </div>
-    <el-table :data="crowds" style="margin-top: 20px">
+
+    <div class="search-bar">
+      <el-input
+        v-model="searchKeyword"
+        placeholder="搜索名称..."
+        clearable
+        style="width: 200px"
+        @input="handleSearch"
+      >
+        <template #prefix>
+          <span>🔍</span>
+        </template>
+      </el-input>
+      <span class="result-count">共 {{ filteredCrowds.length }} 条</span>
+    </div>
+
+    <el-table :data="filteredCrowds" style="margin-top: 16px">
       <el-table-column prop="id" label="ID" width="80" />
       <el-table-column prop="name" label="名称" />
       <el-table-column prop="sort_order" label="排序" width="100" />
@@ -34,15 +50,23 @@
 </template>
 
 <script setup>
-import { ref, onMounted } from 'vue'
+import { ref, computed, onMounted } from 'vue'
 import { getCrowds, createCrowd, updateCrowd, deleteCrowd } from '@/api/category'
 import { ElMessage, ElMessageBox } from 'element-plus'
 
 const crowds = ref([])
 const dialogVisible = ref(false)
 const form = ref({})
+const searchKeyword = ref('')
+
+const filteredCrowds = computed(() => {
+  if (!searchKeyword.value) return crowds.value
+  const kw = searchKeyword.value.toLowerCase()
+  return crowds.value.filter(c => c.name.toLowerCase().includes(kw))
+})
 
 const loadData = async () => { crowds.value = await getCrowds() }
+const handleSearch = () => { /* computed property handles filtering */ }
 const handleEdit = (row) => { form.value = { ...row }; dialogVisible.value = true }
 const handleSave = async () => {
   form.value.id ? await updateCrowd(form.value.id, form.value) : await createCrowd(form.value)
@@ -60,4 +84,14 @@ onMounted(() => loadData())
 
 <style scoped>
 .header { display: flex; justify-content: space-between; align-items: center; }
+.search-bar {
+  display: flex;
+  align-items: center;
+  gap: 12px;
+  margin-top: 16px;
+}
+.result-count {
+  color: #909399;
+  font-size: 13px;
+}
 </style>

admin/src/views/category/Type.vue

@@ -4,7 +4,23 @@
       <h2>类型管理</h2>
       <el-button type="primary" @click="dialogVisible = true; form = {}">添加</el-button>
     </div>
-    <el-table :data="types" style="margin-top: 20px">
+
+    <div class="search-bar">
+      <el-input
+        v-model="searchKeyword"
+        placeholder="搜索名称..."
+        clearable
+        style="width: 200px"
+        @input="handleSearch"
+      >
+        <template #prefix>
+          <span>🔍</span>
+        </template>
+      </el-input>
+      <span class="result-count">共 {{ filteredTypes.length }} 条</span>
+    </div>
+
+    <el-table :data="paginatedTypes" style="margin-top: 16px">
       <el-table-column prop="id" label="ID" width="80" />
       <el-table-column prop="name" label="名称" />
       <el-table-column prop="sort_order" label="排序" width="100" />
@@ -34,15 +50,26 @@
 </template>
 
 <script setup>
-import { ref, onMounted } from 'vue'
+import { ref, computed, onMounted } from 'vue'
 import { getTypes, createType, updateType, deleteType } from '@/api/category'
 import { ElMessage, ElMessageBox } from 'element-plus'
 
 const types = ref([])
 const dialogVisible = ref(false)
 const form = ref({})
+const searchKeyword = ref('')
+const searchResults = ref([])
+
+const filteredTypes = computed(() => {
+  if (!searchKeyword.value) return types.value
+  const kw = searchKeyword.value.toLowerCase()
+  return types.value.filter(t => t.name.toLowerCase().includes(kw))
+})
+
+const paginatedTypes = computed(() => filteredTypes.value)
 
 const loadData = async () => { types.value = await getTypes() }
+const handleSearch = () => { /* computed property handles filtering */ }
 const handleEdit = (row) => { form.value = { ...row }; dialogVisible.value = true }
 const handleSave = async () => {
   form.value.id ? await updateType(form.value.id, form.value) : await createType(form.value)
@@ -60,4 +87,14 @@ onMounted(() => loadData())
 
 <style scoped>
 .header { display: flex; justify-content: space-between; align-items: center; }
+.search-bar {
+  display: flex;
+  align-items: center;
+  gap: 12px;
+  margin-top: 16px;
+}
+.result-count {
+  color: #909399;
+  font-size: 13px;
+}
 </style>

admin/src/views/dashboard/index.vue

@@ -10,14 +10,6 @@
           </div>
         </el-card>
       </el-col>
-      <el-col :span="6">
-        <el-card>
-          <div class="stat-card">
-            <div class="stat-value">{{ stats.pending }}</div>
-            <div class="stat-label">待审核</div>
-          </div>
-        </el-card>
-      </el-col>
       <el-col :span="6">
         <el-card>
           <div class="stat-card">
@@ -26,6 +18,24 @@
           </div>
         </el-card>
       </el-col>
+      <el-col :span="6">
+        <el-card>
+          <div class="stat-card">
+            <div class="stat-value">{{ stats.pending }}</div>
+            <div class="stat-label">待审核</div>
+          </div>
+        </el-card>
+      </el-col>
+      <el-col :span="6">
+        <el-card>
+          <div class="stat-card">
+            <div class="stat-value">{{ stats.rejected }}</div>
+            <div class="stat-label">已拒绝</div>
+          </div>
+        </el-card>
+      </el-col>
+    </el-row>
+    <el-row :gutter="20" style="margin-top: 20px">
       <el-col :span="6">
         <el-card>
           <div class="stat-card">
@@ -34,6 +44,14 @@
           </div>
         </el-card>
       </el-col>
+      <el-col :span="6">
+        <el-card>
+          <div class="stat-card">
+            <div class="stat-value">{{ stats.likes }}</div>
+            <div class="stat-label">总点赞</div>
+          </div>
+        </el-card>
+      </el-col>
     </el-row>
   </div>
 </template>
@@ -42,11 +60,22 @@
 import { ref, onMounted } from 'vue'
 import { getStats } from '@/api/joke'
 
-const stats = ref({ total: 0, pending: 0, approved: 0, views: 0 })
+const stats = ref({ total: 0, pending: 0, approved: 0, rejected: 0, views: 0, likes: 0 })
 
 onMounted(async () => {
-  const res = await getStats()
-  stats.value = res
+  try {
+    const res = await getStats()
+    stats.value = {
+      total: res.total_jokes || 0,
+      pending: res.pending_jokes || 0,
+      approved: res.approved_jokes || 0,
+      rejected: res.rejected_jokes || 0,
+      views: res.total_views || 0,
+      likes: res.total_likes || 0,
+    }
+  } catch (e) {
+    console.error('获取统计数据失败', e)
+  }
 })
 </script>
 

admin/src/views/feedback/List.vue

@@ -0,0 +1,87 @@
+<template>
+  <div class="feedback-list">
+    <div class="header">
+      <h2>反馈管理</h2>
+    </div>
+
+    <div class="search-bar">
+      <el-input
+        v-model="searchKeyword"
+        placeholder="搜索名称/内容..."
+        clearable
+        style="width: 240px"
+      >
+        <template #prefix><span>🔍</span></template>
+      </el-input>
+      <span class="result-count">共 {{ filteredFeedbacks.length }} 条</span>
+    </div>
+
+    <el-table :data="filteredFeedbacks" v-loading="loading" style="margin-top: 16px">
+      <el-table-column prop="id" label="ID" width="60" />
+      <el-table-column prop="name" label="名称" width="120" />
+      <el-table-column prop="email" label="邮箱" width="200" />
+      <el-table-column prop="content" label="内容" min-width="300" show-overflow-tooltip />
+      <el-table-column prop="created_at" label="提交时间" width="180" />
+      <el-table-column label="操作" width="120">
+        <template #default="{ row }">
+          <el-button size="small" type="danger" @click="handleDelete(row.id)">删除</el-button>
+        </template>
+      </el-table-column>
+    </el-table>
+  </div>
+</template>
+
+<script setup>
+import { ref, computed, onMounted } from 'vue'
+import { getFeedbacks, deleteFeedback } from '@/api/feedback'
+import { ElMessage, ElMessageBox } from 'element-plus'
+
+const feedbacks = ref([])
+const loading = ref(false)
+const searchKeyword = ref('')
+
+const filteredFeedbacks = computed(() => {
+  if (!searchKeyword.value) return feedbacks.value
+  const kw = searchKeyword.value.toLowerCase()
+  return feedbacks.value.filter(f =>
+    (f.name && f.name.toLowerCase().includes(kw)) ||
+    (f.content && f.content.toLowerCase().includes(kw))
+  )
+})
+
+const loadFeedbacks = async () => {
+  loading.value = true
+  try {
+    feedbacks.value = await getFeedbacks()
+  } finally {
+    loading.value = false
+  }
+}
+
+const handleDelete = async (id) => {
+  await ElMessageBox.confirm('确定删除?')
+  await deleteFeedback(id)
+  ElMessage.success('删除成功')
+  loadFeedbacks()
+}
+
+onMounted(() => loadFeedbacks())
+</script>
+
+<style scoped>
+.header {
+  display: flex;
+  justify-content: space-between;
+  align-items: center;
+}
+.search-bar {
+  display: flex;
+  align-items: center;
+  gap: 12px;
+  margin-top: 16px;
+}
+.result-count {
+  color: #909399;
+  font-size: 13px;
+}
+</style>

admin/src/views/joke/Edit.vue

@@ -1,20 +1,40 @@
 <template>
   <div class="joke-edit">
     <h2>{{ isEdit ? '编辑笑话' : '添加笑话' }}</h2>
-    <el-form :model="form" ref="formRef" label-width="100px" style="margin-top: 20px; max-width: 600px">
+    <el-form :model="form" ref="formRef" label-width="100px" style="margin-top: 20px; max-width: 800px">
       <el-form-item label="标题" prop="title">
         <el-input v-model="form.title" />
       </el-form-item>
       <el-form-item label="内容" prop="content">
-        <el-input v-model="form.content" type="textarea" :rows="6" />
+        <el-input v-model="form.content" type="textarea" :rows="4" />
       </el-form-item>
-      <el-form-item label="类型" prop="type_id">
-        <el-select v-model="form.type_id" placeholder="选择类型">
+      <el-form-item label="AI润色版本" prop="polished_content">
+        <el-input v-model="form.polished_content" type="textarea" :rows="4" placeholder="AI 润色后的笑话内容" />
+      </el-form-item>
+      <el-row :gutter="20">
+        <el-col :span="8">
+          <el-form-item label="AI评分" prop="ai_score">
+            <el-input-number v-model="form.ai_score" :min="1" :max="10" :step="0.1" precision="1" placeholder="1-10分" />
+          </el-form-item>
+        </el-col>
+        <el-col :span="16">
+          <el-form-item label="AI等级" prop="ai_level">
+            <el-select v-model="form.ai_level" placeholder="AI 质量评价" style="width: 100%">
+              <el-option label="⭐⭐⭐ 精品 (excellent)" value="excellent" />
+              <el-option label="⭐⭐ 良好 (good)" value="good" />
+              <el-option label="⭐ 普通 (ordinary)" value="ordinary" />
+              <el-option label="⚠️ 待优化 (poor)" value="poor" />
+            </el-select>
+          </el-form-item>
+        </el-col>
+      </el-row>
+      <el-form-item label="类型" prop="type_ids">
+        <el-select v-model="form.type_ids" multiple placeholder="选择类型（可多选）">
           <el-option v-for="t in types" :key="t.id" :label="t.name" :value="t.id" />
         </el-select>
       </el-form-item>
-      <el-form-item label="人群" prop="crowd_id">
-        <el-select v-model="form.crowd_id" placeholder="选择人群">
+      <el-form-item label="人群" prop="crowd_ids">
+        <el-select v-model="form.crowd_ids" multiple placeholder="选择人群（可多选）">
           <el-option v-for="c in crowds" :key="c.id" :label="c.name" :value="c.id" />
         </el-select>
       </el-form-item>
@@ -46,7 +66,7 @@ const formRef = ref()
 const isEdit = ref(false)
 
 const form = reactive({
-  title: '', content: '', type_id: null, crowd_id: null, status: 'pending'
+  title: '', content: '', polished_content: '', ai_score: null, ai_level: '', type_ids: [], crowd_ids: [], status: 'pending'
 })
 
 const types = ref([])
@@ -59,7 +79,16 @@ const loadData = async () => {
   if (route.params.id) {
     isEdit.value = true
     const res = await getJoke(route.params.id)
-    Object.assign(form, res)
+    Object.assign(form, {
+      title: res.title,
+      content: res.content,
+      polished_content: res.polished_content || '',
+      ai_score: res.ai_score || null,
+      ai_level: res.ai_level || '',
+      type_ids: res.type_ids || [],
+      crowd_ids: res.crowd_ids || [],
+      status: res.status
+    })
   }
 }
 

admin/src/views/joke/List.vue

@@ -4,11 +4,64 @@
       <h2>笑话管理</h2>
       <el-button type="primary" @click="$router.push('/jokes/edit')">添加笑话</el-button>
     </div>
-    <el-table :data="jokeStore.jokes" v-loading="jokeStore.loading" style="margin-top: 20px">
+
+    <!-- 搜索筛选 -->
+    <div class="search-bar">
+      <el-input
+        v-model="searchKeyword"
+        placeholder="搜索标题/内容..."
+        clearable
+        style="width: 240px"
+        @keyup.enter="handleSearch"
+      >
+        <template #prefix>
+          <span class="search-icon">🔍</span>
+        </template>
+      </el-input>
+      <el-select v-model="filterStatus" placeholder="状态" clearable style="width: 120px" @change="handleSearch">
+        <el-option label="全部" value="" />
+        <el-option label="已发布" value="approved" />
+        <el-option label="待审核" value="pending" />
+        <el-option label="已拒绝" value="rejected" />
+      </el-select>
+      <el-select v-model="filterLevel" placeholder="AI等级" clearable style="width: 130px" @change="handleSearch">
+        <el-option label="全部" value="" />
+        <el-option label="⭐⭐⭐ 精品" value="excellent" />
+        <el-option label="⭐⭐ 良好" value="good" />
+        <el-option label="⭐ 普通" value="ordinary" />
+        <el-option label="⚠️ 待优化" value="poor" />
+      </el-select>
+      <el-button type="primary" @click="handleSearch">搜索</el-button>
+      <el-button @click="resetSearch">重置</el-button>
+    </div>
+
+    <el-table :data="jokeStore.jokes" v-loading="jokeStore.loading" style="margin-top: 16px">
       <el-table-column prop="id" label="ID" width="60" />
-      <el-table-column prop="title" label="标题" />
-      <el-table-column prop="type_name" label="类型" width="100" />
-      <el-table-column prop="crowd_name" label="人群" width="100" />
+      <el-table-column prop="title" label="标题" min-width="150" show-overflow-tooltip />
+      <el-table-column label="AI等级" width="100">
+        <template #default="{ row }">
+          <el-tag v-if="row.ai_level" size="small" :type="getLevelType(row.ai_level)">
+            {{ getLevelText(row.ai_level) }}
+          </el-tag>
+          <span v-else class="text-muted">-</span>
+        </template>
+      </el-table-column>
+      <el-table-column label="润色" width="80">
+        <template #default="{ row }">
+          <el-tag v-if="row.polished_content" type="success" size="small">已润色</el-tag>
+          <span v-else class="text-muted">-</span>
+        </template>
+      </el-table-column>
+      <el-table-column label="类型" width="140">
+        <template #default="{ row }">
+          <el-tag v-for="name in (row.type_names || [])" :key="name" size="small" style="margin:1px">{{ name }}</el-tag>
+        </template>
+      </el-table-column>
+      <el-table-column label="人群" width="140">
+        <template #default="{ row }">
+          <el-tag v-for="name in (row.crowd_names || [])" :key="name" size="small" type="success" style="margin:1px">{{ name }}</el-tag>
+        </template>
+      </el-table-column>
       <el-table-column prop="status" label="状态" width="100">
         <template #default="{ row }">
           <el-tag :type="row.status === 'approved' ? 'success' : row.status === 'pending' ? 'warning' : 'danger'">
@@ -16,8 +69,8 @@
           </el-tag>
         </template>
       </el-table-column>
-      <el-table-column prop="view_count" label="浏览" width="80" />
-      <el-table-column prop="like_count" label="点赞" width="80" />
+      <el-table-column prop="view_count" label="浏览" width="70" />
+      <el-table-column prop="like_count" label="点赞" width="70" />
       <el-table-column label="操作" width="200">
         <template #default="{ row }">
           <el-button size="small" @click="$router.push(`/jokes/edit/${row.id}`)">编辑</el-button>
@@ -25,28 +78,69 @@
         </template>
       </el-table-column>
     </el-table>
-    <el-pagination
-      style="margin-top: 20px"
-      :current-page="page"
-      :page-size="20"
-      :total="jokeStore.total"
-      layout="total, prev, pager, next"
-      @current-change="loadJokes"
-    />
+    <div class="pagination-wrapper">
+      <el-pagination
+        :current-page="page"
+        :page-size="20"
+        :total="jokeStore.total"
+        layout="total, prev, pager, next, jumper"
+        @current-change="loadJokes"
+      />
+      <div class="page-jump">
+        跳至 <el-input-number
+          v-model="jumpPage"
+          :min="1"
+          :max="maxPage"
+          size="small"
+          controls-position="right"
+          @change="handleJump"
+        /> 页
+      </div>
+    </div>
   </div>
 </template>
 
 <script setup>
-import { ref, onMounted } from 'vue'
+import { ref, computed, onMounted } from 'vue'
 import { useJokeStore } from '@/stores/joke'
 import { ElMessage, ElMessageBox } from 'element-plus'
 
 const jokeStore = useJokeStore()
 const page = ref(1)
+const jumpPage = ref(1)
+const searchKeyword = ref('')
+const filterStatus = ref('')
+const filterLevel = ref('')
+
+const maxPage = computed(() => Math.ceil(jokeStore.total / 20) || 1)
 
 const loadJokes = (p = 1) => {
   page.value = p
-  jokeStore.fetchJokes({ page: p, page_size: 20 })
+  jumpPage.value = p
+  jokeStore.fetchJokes({
+    page: p,
+    page_size: 20,
+    keyword: searchKeyword.value || undefined,
+    status: filterStatus.value || undefined,
+    ai_level: filterLevel.value || undefined,
+  })
+}
+
+const handleJump = (val) => {
+  if (val && val >= 1 && val <= maxPage.value) {
+    loadJokes(val)
+  }
+}
+
+const handleSearch = () => {
+  loadJokes(1)
+}
+
+const resetSearch = () => {
+  searchKeyword.value = ''
+  filterStatus.value = ''
+  filterLevel.value = ''
+  loadJokes(1)
 }
 
 const handleDelete = async (id) => {
@@ -56,6 +150,26 @@ const handleDelete = async (id) => {
   loadJokes(page.value)
 }
 
+function getLevelText(level) {
+  const map = {
+    'excellent': '⭐⭐⭐',
+    'good': '⭐⭐',
+    'ordinary': '⭐',
+    'poor': '待优化'
+  }
+  return map[level] || level
+}
+
+function getLevelType(level) {
+  const map = {
+    'excellent': 'warning',
+    'good': 'success',
+    'ordinary': 'info',
+    'poor': 'danger'
+  }
+  return map[level] || 'info'
+}
+
 onMounted(() => loadJokes())
 </script>
 
@@ -65,4 +179,33 @@ onMounted(() => loadJokes())
   justify-content: space-between;
   align-items: center;
 }
+.search-bar {
+  display: flex;
+  gap: 10px;
+  align-items: center;
+  margin-top: 16px;
+  flex-wrap: wrap;
+}
+.search-icon {
+  font-size: 14px;
+}
+.text-muted {
+  color: #999;
+}
+.pagination-wrapper {
+  display: flex;
+  align-items: center;
+  justify-content: space-between;
+  margin-top: 20px;
+}
+.page-jump {
+  display: flex;
+  align-items: center;
+  gap: 8px;
+  color: #606266;
+  font-size: 13px;
+}
+:deep(.el-input-number) {
+  width: 80px;
+}
 </style>

admin/src/views/links/Edit.vue

@@ -0,0 +1,60 @@
+<template>
+  <div class="link-edit">
+    <h2>{{ isEdit ? '编辑友链' : '添加友链' }}</h2>
+    <el-form :model="form" ref="formRef" label-width="120px" style="margin-top: 20px; max-width: 600px">
+      <el-form-item label="名称" prop="name">
+        <el-input v-model="form.name" />
+      </el-form-item>
+      <el-form-item label="URL" prop="url">
+        <el-input v-model="form.url" placeholder="https://" />
+      </el-form-item>
+      <el-form-item label="描述" prop="description">
+        <el-input v-model="form.description" type="textarea" :rows="3" />
+      </el-form-item>
+      <el-form-item label="排序" prop="sort_order">
+        <el-input-number v-model="form.sort_order" :min="0" />
+      </el-form-item>
+      <el-form-item>
+        <el-button type="primary" @click="handleSave">保存</el-button>
+        <el-button @click="$router.back()">取消</el-button>
+      </el-form-item>
+    </el-form>
+  </div>
+</template>
+
+<script setup>
+import { ref, reactive, onMounted } from 'vue'
+import { useRoute, useRouter } from 'vue-router'
+import { getLink, createLink, updateLink } from '@/api/link'
+import { ElMessage } from 'element-plus'
+
+const route = useRoute()
+const router = useRouter()
+const formRef = ref()
+const isEdit = ref(false)
+
+const form = reactive({
+  name: '', url: '', description: '', sort_order: 0
+})
+
+const loadData = async () => {
+  if (route.params.id) {
+    isEdit.value = true
+    const res = await getLink(route.params.id)
+    Object.assign(form, res)
+  }
+}
+
+const handleSave = async () => {
+  if (isEdit.value) {
+    await updateLink(route.params.id, form)
+    ElMessage.success('更新成功')
+  } else {
+    await createLink(form)
+    ElMessage.success('创建成功')
+  }
+  router.push('/links')
+}
+
+onMounted(() => loadData())
+</script>

admin/src/views/links/List.vue

@@ -0,0 +1,93 @@
+<template>
+  <div class="link-list">
+    <div class="header">
+      <h2>友链管理</h2>
+      <el-button type="primary" @click="$router.push('/links/edit')">添加友链</el-button>
+    </div>
+
+    <div class="search-bar">
+      <el-input
+        v-model="searchKeyword"
+        placeholder="搜索名称/URL..."
+        clearable
+        style="width: 240px"
+      >
+        <template #prefix><span>🔍</span></template>
+      </el-input>
+      <span class="result-count">共 {{ filteredLinks.length }} 条</span>
+    </div>
+
+    <el-table :data="filteredLinks" v-loading="loading" style="margin-top: 16px">
+      <el-table-column prop="id" label="ID" width="60" />
+      <el-table-column prop="name" label="名称" />
+      <el-table-column prop="url" label="URL" min-width="200">
+        <template #default="{ row }">
+          <a :href="row.url" target="_blank">{{ row.url }}</a>
+        </template>
+      </el-table-column>
+      <el-table-column prop="description" label="描述" min-width="200" />
+      <el-table-column prop="sort_order" label="排序" width="80" />
+      <el-table-column prop="created_at" label="创建时间" width="180" />
+      <el-table-column label="操作" width="200">
+        <template #default="{ row }">
+          <el-button size="small" @click="$router.push(`/links/edit/${row.id}`)">编辑</el-button>
+          <el-button size="small" type="danger" @click="handleDelete(row.id)">删除</el-button>
+        </template>
+      </el-table-column>
+    </el-table>
+  </div>
+</template>
+
+<script setup>
+import { ref, computed, onMounted } from 'vue'
+import { getLinks, deleteLink } from '@/api/link'
+import { ElMessage, ElMessageBox } from 'element-plus'
+
+const links = ref([])
+const loading = ref(false)
+const searchKeyword = ref('')
+
+const filteredLinks = computed(() => {
+  if (!searchKeyword.value) return links.value
+  const kw = searchKeyword.value.toLowerCase()
+  return links.value.filter(l =>
+    l.name.toLowerCase().includes(kw) || l.url.toLowerCase().includes(kw)
+  )
+})
+
+const loadLinks = async () => {
+  loading.value = true
+  try {
+    links.value = await getLinks()
+  } finally {
+    loading.value = false
+  }
+}
+
+const handleDelete = async (id) => {
+  await ElMessageBox.confirm('确定删除?')
+  await deleteLink(id)
+  ElMessage.success('删除成功')
+  loadLinks()
+}
+
+onMounted(() => loadLinks())
+</script>
+
+<style scoped>
+.header {
+  display: flex;
+  justify-content: space-between;
+  align-items: center;
+}
+.search-bar {
+  display: flex;
+  align-items: center;
+  gap: 12px;
+  margin-top: 16px;
+}
+.result-count {
+  color: #909399;
+  font-size: 13px;
+}
+</style>

admin/src/views/setting/Index.vue

@@ -0,0 +1,300 @@
+<template>
+  <div class="setting-page">
+    <h2>AI 配置</h2>
+
+    <!-- AI 基础配置 -->
+    <el-card style="margin-top: 20px">
+      <template #header>
+        <span>API 基础配置</span>
+      </template>
+      <el-form :model="form" label-width="120px">
+        <el-form-item label="提供商">
+          <el-select v-model="form.provider" style="width: 200px">
+            <el-option label="NVIDIA NIM" value="nvidia" />
+            <el-option label="OpenAI" value="openai" />
+            <el-option label="Anthropic" value="anthropic" />
+          </el-select>
+        </el-form-item>
+        <el-form-item label="API 地址">
+          <el-input v-model="form.api_base" placeholder="https://integrate.api.nvidia.com/v1" style="width: 400px" />
+        </el-form-item>
+        <el-form-item label="API 密钥">
+          <el-input v-model="form.api_key" type="password" show-password style="width: 400px" placeholder="nvapi-..." />
+        </el-form-item>
+        <el-form-item label="模型名称">
+          <el-input v-model="form.model_name" style="width: 300px" placeholder="nvidia/llama-3.1-nemotron-70b-instruct" />
+        </el-form-item>
+        <el-form-item label="温度">
+          <el-input-number v-model="form.temperature" :min="0" :max="2" :step="0.1" />
+        </el-form-item>
+        <el-form-item label="最大 Token">
+          <el-input-number v-model="form.max_tokens" :min="1" :max="8192" />
+        </el-form-item>
+      </el-form>
+    </el-card>
+
+    <!-- 爬虫配置 -->
+    <el-card style="margin-top: 20px">
+      <template #header>
+        <span>爬虫配置</span>
+      </template>
+      <el-form :model="form" label-width="120px">
+        <el-form-item label="启用爬虫">
+          <el-switch v-model="form.crawl_enabled" />
+        </el-form-item>
+        <el-form-item label="搜索关键词">
+          <el-input
+            v-model="form.crawl_keywords"
+            type="textarea"
+            :rows="2"
+            style="width: 400px"
+            placeholder="多个关键词用逗号分隔，如：冷笑话,段子,谐音梗"
+          />
+        </el-form-item>
+        <el-form-item label="每轮页数">
+          <el-input-number v-model="form.max_pages_per_run" :min="1" :max="20" />
+        </el-form-item>
+      </el-form>
+    </el-card>
+
+    <!-- 提示词配置 -->
+    <el-card style="margin-top: 20px">
+      <template #header>
+        <span>提示词模板</span>
+      </template>
+      <el-form :model="form" label-width="140px">
+        <el-form-item label="AI 生成笑话">
+          <el-input
+            v-model="form.generate_prompt"
+            type="textarea"
+            :rows="6"
+            placeholder="使用 {context}/{style}/{length_requirement} 作为占位符"
+            style="width: 100%"
+          />
+        </el-form-item>
+        <el-divider />
+
+        <el-form-item label="优化器-质量检测">
+          <el-input
+            v-model="form.optimizer_quality_prompt"
+            type="textarea"
+            :rows="4"
+            style="width: 100%"
+          />
+        </el-form-item>
+        <el-form-item label="质量检测温度">
+          <el-input-number v-model="form.optimizer_quality_temperature" :min="0" :max="2" :step="0.1" />
+        </el-form-item>
+        <el-divider />
+
+        <el-form-item label="优化器-润色">
+          <el-input
+            v-model="form.optimizer_polish_prompt"
+            type="textarea"
+            :rows="4"
+            style="width: 100%"
+          />
+        </el-form-item>
+        <el-form-item label="润色温度">
+          <el-input-number v-model="form.optimizer_polish_temperature" :min="0" :max="2" :step="0.1" />
+        </el-form-item>
+        <el-divider />
+
+        <el-form-item label="优化器-评价分类">
+          <el-input
+            v-model="form.optimizer_evaluate_prompt"
+            type="textarea"
+            :rows="4"
+            style="width: 100%"
+          />
+        </el-form-item>
+        <el-form-item label="评价温度">
+          <el-input-number v-model="form.optimizer_evaluate_temperature" :min="0" :max="2" :step="0.1" />
+        </el-form-item>
+        <el-divider />
+
+        <el-form-item label="爬虫-提取笑话">
+          <el-input
+            v-model="form.crawler_extract_prompt"
+            type="textarea"
+            :rows="4"
+            style="width: 100%"
+          />
+        </el-form-item>
+        <el-form-item label="爬虫-改写">
+          <el-input
+            v-model="form.crawler_rewrite_prompt"
+            type="textarea"
+            :rows="4"
+            style="width: 100%"
+          />
+        </el-form-item>
+      </el-form>
+    </el-card>
+
+    <div style="margin-top: 20px">
+      <el-button type="primary" @click="handleSave" :loading="saving">保存配置</el-button>
+    </div>
+
+    <!-- 历史配置 -->
+    <el-card style="margin-top: 20px">
+      <template #header>
+        <span>历史配置</span>
+      </template>
+      <el-table :data="allSettings" style="width: 100%">
+        <el-table-column prop="id" label="ID" width="60" />
+        <el-table-column prop="provider" label="提供商" width="100" />
+        <el-table-column prop="model_name" label="模型" />
+        <el-table-column prop="api_base" label="API 地址" show-overflow-tooltip />
+        <el-table-column prop="is_active" label="激活" width="80">
+          <template #default="{ row }">
+            <el-tag :type="row.is_active ? 'success' : 'info'" size="small">
+              {{ row.is_active ? '是' : '否' }}
+            </el-tag>
+          </template>
+        </el-table-column>
+        <el-table-column prop="created_at" label="创建时间" width="180" />
+        <el-table-column label="操作" width="200">
+          <template #default="{ row }">
+            <el-button size="small" @click="handleEdit(row)">编辑</el-button>
+            <el-button size="small" type="warning" @click="handleToggle(row)" v-if="!row.is_active">激活</el-button>
+            <el-button size="small" type="danger" @click="handleDelete(row.id)">删除</el-button>
+          </template>
+        </el-table-column>
+      </el-table>
+    </el-card>
+  </div>
+</template>
+
+<script setup>
+import { ref, onMounted } from 'vue'
+import { ElMessage, ElMessageBox } from 'element-plus'
+import { getSettings, getActiveSetting, createSetting, updateSetting, toggleSetting, deleteSetting } from '@/api/setting'
+
+const form = ref({
+  provider: 'nvidia',
+  api_base: 'https://integrate.api.nvidia.com/v1',
+  api_key: '',
+  model_name: 'nvidia/llama-3.1-nemotron-70b-instruct',
+  temperature: 0.7,
+  max_tokens: 2048,
+  crawl_enabled: false,
+  crawl_keywords: '冷笑话,段子,谐音梗',
+  max_pages_per_run: 3,
+  generate_prompt: '',
+  optimizer_quality_prompt: '',
+  optimizer_polish_prompt: '',
+  optimizer_evaluate_prompt: '',
+  optimizer_quality_temperature: 0.3,
+  optimizer_polish_temperature: 0.8,
+  optimizer_evaluate_temperature: 0.3,
+  crawler_extract_prompt: '',
+  crawler_rewrite_prompt: '',
+})
+
+const allSettings = ref([])
+const saving = ref(false)
+const editingId = ref(null)
+
+const loadData = async () => {
+  try {
+    const active = await getActiveSetting()
+    editingId.value = active.id
+    form.value = {
+      provider: active.provider,
+      api_base: active.api_base,
+      api_key: active.api_key,
+      model_name: active.model_name,
+      temperature: active.temperature,
+      max_tokens: active.max_tokens,
+      crawl_enabled: active.crawl_enabled,
+      crawl_keywords: active.crawl_keywords || '',
+      max_pages_per_run: active.max_pages_per_run,
+      generate_prompt: active.generate_prompt || '',
+      optimizer_quality_prompt: active.optimizer_quality_prompt || '',
+      optimizer_polish_prompt: active.optimizer_polish_prompt || '',
+      optimizer_evaluate_prompt: active.optimizer_evaluate_prompt || '',
+      optimizer_quality_temperature: active.optimizer_quality_temperature ?? 0.3,
+      optimizer_polish_temperature: active.optimizer_polish_temperature ?? 0.8,
+      optimizer_evaluate_temperature: active.optimizer_evaluate_temperature ?? 0.3,
+      crawler_extract_prompt: active.crawler_extract_prompt || '',
+      crawler_rewrite_prompt: active.crawler_rewrite_prompt || '',
+    }
+  } catch (e) {
+    // 没有激活配置，使用默认值
+  }
+
+  try {
+    allSettings.value = await getSettings()
+  } catch (e) {
+    console.error(e)
+  }
+}
+
+const handleSave = async () => {
+  saving.value = true
+  try {
+    if (editingId.value) {
+      await updateSetting(editingId.value, form.value)
+      ElMessage.success('保存成功')
+    } else {
+      const created = await createSetting(form.value)
+      editingId.value = created.id
+      await toggleSetting(created.id)
+      ElMessage.success('创建并激活成功')
+    }
+    await loadData()
+  } catch (e) {
+    ElMessage.error('保存失败: ' + (e.message || ''))
+  } finally {
+    saving.value = false
+  }
+}
+
+const handleEdit = (row) => {
+  editingId.value = row.id
+  form.value = {
+    provider: row.provider,
+    api_base: row.api_base,
+    api_key: row.api_key,
+    model_name: row.model_name,
+    temperature: row.temperature,
+    max_tokens: row.max_tokens,
+    crawl_enabled: row.crawl_enabled,
+    crawl_keywords: row.crawl_keywords || '',
+    max_pages_per_run: row.max_pages_per_run,
+    generate_prompt: row.generate_prompt || '',
+    optimizer_quality_prompt: row.optimizer_quality_prompt || '',
+    optimizer_polish_prompt: row.optimizer_polish_prompt || '',
+    optimizer_evaluate_prompt: row.optimizer_evaluate_prompt || '',
+    optimizer_quality_temperature: row.optimizer_quality_temperature ?? 0.3,
+    optimizer_polish_temperature: row.optimizer_polish_temperature ?? 0.8,
+    optimizer_evaluate_temperature: row.optimizer_evaluate_temperature ?? 0.3,
+    crawler_extract_prompt: row.crawler_extract_prompt || '',
+    crawler_rewrite_prompt: row.crawler_rewrite_prompt || '',
+  }
+}
+
+const handleToggle = async (row) => {
+  await ElMessageBox.confirm('确定激活此配置？')
+  await toggleSetting(row.id)
+  ElMessage.success('激活成功')
+  await loadData()
+}
+
+const handleDelete = async (id) => {
+  await ElMessageBox.confirm('确定删除此配置？')
+  await deleteSetting(id)
+  ElMessage.success('删除成功')
+  if (editingId.value === id) {
+    editingId.value = null
+  }
+  await loadData()
+}
+
+onMounted(() => loadData())
+</script>
+
+<style scoped>
+.setting-page { max-width: 900px }
+</style>

admin/src/views/user/List.vue

@@ -0,0 +1,118 @@
+<template>
+  <div class="user-list">
+    <div class="header">
+      <h2>用户管理</h2>
+      <el-button type="primary" @click="showDialog = true; form = {}">添加用户</el-button>
+    </div>
+
+    <el-table :data="users" v-loading="loading" style="margin-top: 16px">
+      <el-table-column prop="id" label="ID" width="60" />
+      <el-table-column prop="username" label="用户名" />
+      <el-table-column label="角色" width="120">
+        <template #default="{ row }">
+          <el-tag :type="row.role === 'admin' ? 'danger' : 'success'" size="small">
+            {{ row.role === 'admin' ? '管理员' : '提交者' }}
+          </el-tag>
+        </template>
+      </el-table-column>
+      <el-table-column prop="created_at" label="创建时间" width="180" />
+      <el-table-column label="操作" width="120">
+        <template #default="{ row }">
+          <el-button size="small" type="danger" @click="handleDelete(row.id)"
+            :disabled="row.id === currentUserId">删除</el-button>
+        </template>
+      </el-table-column>
+    </el-table>
+
+    <!-- 添加用户弹窗 -->
+    <el-dialog v-model="showDialog" :title="form.id ? '编辑用户' : '添加用户'">
+      <el-form :model="form" label-width="80px">
+        <el-form-item label="用户名" required>
+          <el-input v-model="form.username" placeholder="输入用户名" />
+        </el-form-item>
+        <el-form-item label="密码" required>
+          <el-input v-model="form.password" type="password" placeholder="输入密码" show-password />
+        </el-form-item>
+        <el-form-item label="角色">
+          <el-select v-model="form.role" style="width: 100%">
+            <el-option label="管理员 (admin)" value="admin" />
+            <el-option label="提交者 (submitter)" value="submitter" />
+          </el-select>
+        </el-form-item>
+      </el-form>
+      <template #footer>
+        <el-button @click="showDialog = false">取消</el-button>
+        <el-button type="primary" @click="handleSave" :loading="saving">保存</el-button>
+      </template>
+    </el-dialog>
+  </div>
+</template>
+
+<script setup>
+import { ref, onMounted } from 'vue'
+import { getUsers, createUser, deleteUser } from '@/api/user'
+import { ElMessage, ElMessageBox } from 'element-plus'
+
+const users = ref([])
+const loading = ref(false)
+const showDialog = ref(false)
+const saving = ref(false)
+const form = ref({ role: 'submitter' })
+
+// 当前登录用户的 ID（从 token 解析）
+const currentUserId = ref(1) // 假设登录用户是 admin
+
+const loadUsers = async () => {
+  loading.value = true
+  try {
+    users.value = await getUsers()
+    // 从 localStorage 的 token 解析用户 ID（如果有）
+    const token = localStorage.getItem('token')
+    if (token) {
+      try {
+        const payload = JSON.parse(atob(token.split('.')[1]))
+        currentUserId.value = parseInt(payload.sub)
+      } catch (e) {}
+    }
+  } catch (e) {
+    ElMessage.error('获取用户列表失败')
+  } finally {
+    loading.value = false
+  }
+}
+
+const handleSave = async () => {
+  if (!form.value.username || !form.value.password) {
+    ElMessage.warning('请填写用户名和密码')
+    return
+  }
+  saving.value = true
+  try {
+    await createUser(form.value)
+    ElMessage.success('添加成功')
+    showDialog.value = false
+    loadUsers()
+  } catch (e) {
+    ElMessage.error(e.response?.data?.detail || '添加失败')
+  } finally {
+    saving.value = false
+  }
+}
+
+const handleDelete = async (id) => {
+  await ElMessageBox.confirm('确定删除此用户？')
+  await deleteUser(id)
+  ElMessage.success('删除成功')
+  loadUsers()
+}
+
+onMounted(() => loadUsers())
+</script>
+
+<style scoped>
+.header {
+  display: flex;
+  justify-content: space-between;
+  align-items: center;
+}
+</style>

admin/vite.config.js

@@ -9,11 +9,12 @@ export default defineConfig({
       '@': path.resolve(__dirname, './src')
     }
   },
+  base: '/admin/',
   server: {
     port: 3001,
     proxy: {
       '/api': {
-        target: 'http://localhost:8000',
+        target: 'http://localhost:8001',
         changeOrigin: true
       }
     }

api/API操作文档.md

@@ -0,0 +1,125 @@
+# API 操作文档 - 笑话管理系统
+
+## 基础信息
+
+- **API 地址**: `http://<服务器IP>:8001`
+- **默认管理员**: `admin` / `admin123`
+- **认证方式**: JWT Bearer Token
+
+---
+
+## 一、登录获取 Token
+
+```bash
+curl -X POST http://<服务器IP>:8001/api/auth/login \
+  -H "Content-Type: application/json" \
+  -d '{"username": "admin", "password": "admin123"}'
+```
+
+**返回示例**：
+```json
+{
+  "access_token": "eyJhbGciOiJIUzI1NiIs...",
+  "token_type": "bearer"
+}
+```
+
+---
+
+## 二、提交笑话
+
+```bash
+curl -X POST http://<服务器IP>:8001/api/admin/jokes \
+  -H "Content-Type: application/json" \
+  -H "Authorization: Bearer <TOKEN>" \
+  -d '{
+    "title": "笑话标题",
+    "content": "笑话内容正文",
+    "type_id": 1,
+    "crowd_id": 2
+  }'
+```
+
+| 字段 | 类型 | 必填 | 说明 |
+|------|------|------|------|
+| `title` | string | 是 | 笑话标题 |
+| `content` | string | 是 | 笑话正文 |
+| `type_id` | int | 否 | 类型 ID（查分类列表获取） |
+| `crowd_id` | int | 否 | 人群 ID（查分类列表获取） |
+
+提交后 status 默认为 `pending`（待审核）。
+
+---
+
+## 三、查询分类 ID
+
+```bash
+# 查看所有类型
+curl http://<服务器IP>:8001/api/categories/types
+
+# 查看所有人群
+curl http://<服务器IP>:8001/api/categories/crowds
+```
+
+---
+
+## 四、完整脚本示例（批量提交）
+
+```bash
+#!/bin/bash
+SERVER="http://192.168.1.10:8001"
+
+# 登录获取 Token
+TOKEN=$(curl -s -X POST "$SERVER/api/auth/login" \
+  -H "Content-Type: application/json" \
+  -d '{"username":"admin","password":"admin123"}' \
+  | python -c "import sys,json; print(json.load(sys.stdin)['access_token'])")
+
+echo "Token: $TOKEN"
+
+# 批量提交笑话
+jokes=(
+  '{"title":"笑话1","content":"内容1","type_id":1,"crowd_id":2}'
+  '{"title":"笑话2","content":"内容2","type_id":1,"crowd_id":null}'
+)
+
+for joke in "${jokes[@]}"; do
+  curl -X POST "$SERVER/api/admin/jokes" \
+    -H "Content-Type: application/json" \
+    -H "Authorization: Bearer $TOKEN" \
+    -d "$joke"
+  echo ""
+done
+```
+
+---
+
+## 五、其他管理接口
+
+### 查询笑话列表
+```bash
+curl -H "Authorization: Bearer <TOKEN>" \
+  "http://<服务器IP>:8001/api/admin/jokes?page=1&page_size=20&status=pending"
+```
+
+### 审核通过
+```bash
+curl -X PUT http://<服务器IP>:8001/api/admin/jokes/batch-approve \
+  -H "Content-Type: application/json" \
+  -H "Authorization: Bearer <TOKEN>" \
+  -d '[1, 2, 3]'
+```
+
+### 删除笑话
+```bash
+curl -X DELETE http://<服务器IP>:8001/api/admin/jokes/1 \
+  -H "Authorization: Bearer <TOKEN>"
+```
+
+---
+
+## 六、注意事项
+
+1. 将 `<服务器IP>` 替换为实际的服务器 IP 地址
+2. 确保服务器防火墙已放行 **8001** 端口
+3. Token 有过期时间，过期后需重新登录获取

api/app/models/__init__.py

@@ -1,5 +1,8 @@
 from app.models.joke import Joke
 from app.models.category import JokeType, JokeCrowd
 from app.models.user import AdminUser
+from app.models.setting import AiSetting
+from app.models.link import Link
+from app.models.feedback import Feedback
 
-__all__ = ["Joke", "JokeType", "JokeCrowd", "AdminUser"]
+__all__ = ["Joke", "JokeType", "JokeCrowd", "AdminUser", "AiSetting", "Link", "Feedback"]

api/app/models/feedback.py

@@ -0,0 +1,13 @@
+from sqlalchemy import Column, Integer, String, Text, DateTime
+from sqlalchemy.sql import func
+
+from app.database import Base
+
+
+class Feedback(Base):
+    __tablename__ = "feedbacks"
+    id = Column(Integer, primary_key=True, autoincrement=True)
+    name = Column(String(50), nullable=True)
+    email = Column(String(100), nullable=True)
+    content = Column(Text, nullable=False)
+    created_at = Column(DateTime, default=func.now())

api/app/models/joke.py

@@ -1,4 +1,4 @@
-from sqlalchemy import Column, Integer, String, Text, DateTime, ForeignKey
+from sqlalchemy import Column, Integer, String, Text, DateTime, Float, ForeignKey
 from sqlalchemy.orm import relationship
 from sqlalchemy.sql import func
 
@@ -12,11 +12,20 @@ class Joke(Base):
     id = Column(Integer, primary_key=True, autoincrement=True)
     title = Column(String(200), nullable=False)
     content = Column(Text, nullable=False)
+    # AI 润色后的内容
+    polished_content = Column(Text, nullable=True)
+    # AI 评分 (1-10)
+    ai_score = Column(Float, nullable=True)
+    # AI 质量等级: excellent/good/ordinary/poor
+    ai_level = Column(String(20), nullable=True)
+    type_ids = Column(Text, nullable=True)
+    crowd_ids = Column(Text, nullable=True)
     type_id = Column(Integer, ForeignKey("joke_types.id"), nullable=True)
     crowd_id = Column(Integer, ForeignKey("joke_crowds.id"), nullable=True)
     status = Column(String(20), default="pending")
     view_count = Column(Integer, default=0)
     like_count = Column(Integer, default=0)
+    dislike_count = Column(Integer, default=0)
     created_at = Column(DateTime, default=func.now())
     updated_at = Column(DateTime, default=func.now(), onupdate=func.now())
 

api/app/models/link.py

@@ -0,0 +1,16 @@
+from sqlalchemy import Column, Integer, String, DateTime
+from sqlalchemy.sql import func
+
+from app.database import Base
+
+
+class Link(Base):
+    __tablename__ = "links"
+    id = Column(Integer, primary_key=True, autoincrement=True)
+    name = Column(String(100), nullable=False)
+    url = Column(String(500), nullable=False)
+    description = Column(String(200), nullable=True)
+    sort_order = Column(Integer, default=0)
+    status = Column(String(20), default="approved")  # approved=已通过, pending=待审核
+    contact = Column(String(100), nullable=True)       # 申请联系方式
+    created_at = Column(DateTime, default=func.now())

api/app/models/setting.py

@@ -0,0 +1,36 @@
+from sqlalchemy import Boolean, Column, DateTime, Float, Integer, String, Text, func
+
+from app.database import Base
+
+
+class AiSetting(Base):
+    __tablename__ = "ai_settings"
+
+    id = Column(Integer, primary_key=True, autoincrement=True)
+    provider = Column(String(50), nullable=False, default="nvidia")
+    api_base = Column(String(500), nullable=False, default="https://integrate.api.nvidia.com/v1")
+    api_key = Column(String(500), nullable=False, default="")
+    model_name = Column(String(100), nullable=False, default="nvidia/llama-3.1-nemotron-70b-instruct")
+    temperature = Column(Float, nullable=False, default=0.7)
+    max_tokens = Column(Integer, nullable=False, default=2048)
+
+    # 提示词模板
+    generate_prompt = Column(Text, nullable=True, comment="AI 笑话生成提示词")
+    optimizer_quality_prompt = Column(Text, nullable=True, comment="优化器-质量检测提示词")
+    optimizer_polish_prompt = Column(Text, nullable=True, comment="优化器-润色提示词")
+    optimizer_evaluate_prompt = Column(Text, nullable=True, comment="优化器-评价分类提示词")
+    crawler_extract_prompt = Column(Text, nullable=True, comment="爬虫-笑话提取提示词")
+    crawler_rewrite_prompt = Column(Text, nullable=True, comment="爬虫-改写提示词")
+
+    # 优化器各阶段温度
+    optimizer_quality_temperature = Column(Float, nullable=False, default=0.3)
+    optimizer_polish_temperature = Column(Float, nullable=False, default=0.8)
+    optimizer_evaluate_temperature = Column(Float, nullable=False, default=0.3)
+
+    crawl_enabled = Column(Boolean, nullable=False, default=False)
+    crawl_keywords = Column(Text, nullable=True, default="")
+    max_pages_per_run = Column(Integer, nullable=False, default=3)
+
+    is_active = Column(Boolean, nullable=False, default=True)
+    created_at = Column(DateTime, default=func.now())
+    updated_at = Column(DateTime, default=func.now(), onupdate=func.now())

api/app/models/user.py

@@ -5,10 +5,11 @@ from app.database import Base
 
 
 class AdminUser(Base):
-    """管理员用户表模型"""
+    """用户表 - 支持角色：admin(管理员), submitter(提交者)"""
     __tablename__ = "admin_users"
 
     id = Column(Integer, primary_key=True, autoincrement=True)
     username = Column(String(50), unique=True, nullable=False)
     password_hash = Column(String(255), nullable=False)
+    role = Column(String(20), default="submitter")          # admin=管理员, submitter=提交者
     created_at = Column(DateTime, default=func.now())

api/app/routers/__init__.py

@@ -2,5 +2,9 @@ from .jokes import router as jokes_router
 from .categories import router as categories_router
 from .auth import router as auth_router
 from .admin import router as admin_router
+from .settings import router as settings_router
+from .links import router as links_router
+from .feedback import router as feedback_router
+from .generate import router as generate_router
 
-__all__ = ["jokes_router", "categories_router", "auth_router", "admin_router"]
+__all__ = ["jokes_router", "categories_router", "auth_router", "admin_router", "settings_router", "links_router", "feedback_router", "generate_router"]

api/app/routers/__init__.py.bak

@@ -0,0 +1,7 @@
+from .jokes import router as jokes_router
+from .categories import router as categories_router
+from .auth import router as auth_router
+from .admin import router as admin_router
+from .settings import router as settings_router
+
+__all__ = ["jokes_router", "categories_router", "auth_router", "admin_router", "settings_router"]

api/app/routers/admin.py

@@ -1,3 +1,5 @@
+import json
+
 from fastapi import APIRouter, Depends, HTTPException, Header
 from jose import JWTError, jwt
 from sqlalchemy import func
@@ -8,8 +10,14 @@ from app.database import get_db
 from app.models.joke import Joke
 from app.models.category import JokeCrowd, JokeType
 from app.models.user import AdminUser
+from app.models.link import Link
+from app.models.feedback import Feedback
 from app.schemas.category import JokeCrowdCreate, JokeCrowdResponse, JokeTypeCreate, JokeTypeResponse
 from app.schemas.joke import JokeCreate, JokeResponse, JokeUpdate, PaginatedJokeResponse
+from app.schemas.link import LinkCreate, LinkResponse
+from app.schemas.feedback import FeedbackResponse
+from app.schemas.user import UserCreate, UserResponse
+from app.routers.auth import hash_password
 
 router = APIRouter(prefix="/admin", tags=["管理后台"])
 
@@ -39,21 +47,51 @@ def get_current_admin_user(
     return user
 
 
-def joke_to_response(joke: Joke) -> JokeResponse:
+def _parse_ids(raw) -> list[int]:
+    """解析 DB 中的 JSON 列表字符串为 Python list"""
+    if not raw:
+        return []
+    if isinstance(raw, list):
+        return raw
+    try:
+        return json.loads(raw)
+    except Exception:
+        return []
+
+
+def _get_type_names(db: Session, ids: list[int]) -> list[str]:
+    rows = db.query(JokeType).filter(JokeType.id.in_(ids)).all()
+    return [r.name for r in rows]
+
+
+def _get_crowd_names(db: Session, ids: list[int]) -> list[str]:
+    rows = db.query(JokeCrowd).filter(JokeCrowd.id.in_(ids)).all()
+    return [r.name for r in rows]
+
+
+def joke_to_response(joke: Joke, db: Session = None) -> JokeResponse:
     """Convert Joke model to JokeResponse schema."""
+    ids = _parse_ids(joke.type_ids)
+    crowd_ids = _parse_ids(joke.crowd_ids)
+    type_names = _get_type_names(db, ids) if db else []
+    crowd_names = _get_crowd_names(db, crowd_ids) if db else []
     return JokeResponse(
         id=joke.id,
         title=joke.title,
         content=joke.content,
-        type_id=joke.type_id,
-        crowd_id=joke.crowd_id,
+        polished_content=joke.polished_content,
+        ai_score=joke.ai_score,
+        ai_level=joke.ai_level,
+        type_ids=ids,
+        crowd_ids=crowd_ids,
         status=joke.status,
         view_count=joke.view_count,
         like_count=joke.like_count,
+        dislike_count=joke.dislike_count,
         created_at=joke.created_at,
         updated_at=joke.updated_at,
-        type_name=joke.type.name if joke.type else None,
-        crowd_name=joke.crowd.name if joke.crowd else None,
+        type_names=type_names,
+        crowd_names=crowd_names,
     )
 
 
@@ -63,18 +101,29 @@ def admin_list_jokes(
     page: int = 1,
     page_size: int = 20,
     status: str = None,
+    keyword: str = None,
+    ai_level: str = None,
     db: Session = Depends(get_db),
     current_user: AdminUser = Depends(get_current_admin_user),
 ):
-    """获取所有笑话（支持状态筛选）"""
+    """获取所有笑话（支持状态筛选、关键词搜索）"""
+    # 限制 page_size 防止 DoS
+    page_size = max(1, min(page_size, 100))
     query = db.query(Joke)
     if status:
         query = query.filter(Joke.status == status)
+    if keyword:
+        keyword_filter = f"%{keyword}%"
+        query = query.filter(
+            (Joke.title.like(keyword_filter)) | (Joke.content.like(keyword_filter))
+        )
+    if ai_level:
+        query = query.filter(Joke.ai_level == ai_level)
     total = query.count()
     offset = (page - 1) * page_size
     jokes = query.order_by(Joke.created_at.desc()).offset(offset).limit(page_size).all()
     return PaginatedJokeResponse(
-        items=[joke_to_response(j) for j in jokes],
+        items=[joke_to_response(j, db) for j in jokes],
         total=total,
         page=page,
         page_size=page_size,
@@ -88,11 +137,29 @@ def admin_create_joke(
     current_user: AdminUser = Depends(get_current_admin_user),
 ):
     """创建笑话"""
-    db_joke = Joke(**joke.model_dump())
+    data = joke.model_dump()
+    if data.get("type_ids") is not None:
+        data["type_ids"] = json.dumps(data["type_ids"])
+    if data.get("crowd_ids") is not None:
+        data["crowd_ids"] = json.dumps(data["crowd_ids"])
+    db_joke = Joke(**data)
     db.add(db_joke)
     db.commit()
     db.refresh(db_joke)
-    return joke_to_response(db_joke)
+    return joke_to_response(db_joke, db)
+
+
+@router.get("/jokes/{joke_id}", response_model=JokeResponse)
+def admin_get_joke(
+    joke_id: int,
+    db: Session = Depends(get_db),
+    current_user: AdminUser = Depends(get_current_admin_user),
+):
+    """获取单个笑话"""
+    joke = db.query(Joke).filter(Joke.id == joke_id).first()
+    if not joke:
+        raise HTTPException(status_code=404, detail="笑话不存在")
+    return joke_to_response(joke, db)
 
 
 @router.put("/jokes/{joke_id}", response_model=JokeResponse)
@@ -106,12 +173,15 @@ def admin_update_joke(
     db_joke = db.query(Joke).filter(Joke.id == joke_id).first()
     if not db_joke:
         raise HTTPException(status_code=404, detail="笑话不存在")
+
     update_data = joke.model_dump(exclude_unset=True)
     for key, value in update_data.items():
+        if key in ("type_ids", "crowd_ids") and value is not None:
+            value = json.dumps(value)
         setattr(db_joke, key, value)
     db.commit()
     db.refresh(db_joke)
-    return joke_to_response(db_joke)
+    return joke_to_response(db_joke, db)
 
 
 @router.delete("/jokes/{joke_id}")
@@ -152,14 +222,18 @@ def admin_stats(
     total_jokes = db.query(Joke).count()
     approved_jokes = db.query(Joke).filter(Joke.status == "approved").count()
     pending_jokes = db.query(Joke).filter(Joke.status == "pending").count()
+    rejected_jokes = db.query(Joke).filter(Joke.status == "rejected").count()
     total_views = db.query(Joke).with_entities(func.sum(Joke.view_count)).scalar() or 0
     total_likes = db.query(Joke).with_entities(func.sum(Joke.like_count)).scalar() or 0
+    total_dislikes = db.query(Joke).with_entities(func.sum(Joke.dislike_count)).scalar() or 0
     return {
         "total_jokes": total_jokes,
         "approved_jokes": approved_jokes,
         "pending_jokes": pending_jokes,
+        "rejected_jokes": rejected_jokes,
         "total_views": total_views,
         "total_likes": total_likes,
+        "total_dislikes": total_dislikes,
     }
 
 
@@ -209,8 +283,11 @@ def admin_delete_type(
     db_type = db.query(JokeType).filter(JokeType.id == type_id).first()
     if not db_type:
         raise HTTPException(status_code=404, detail="类型不存在")
-    # Check if there are jokes using this type
-    joke_count = db.query(Joke).filter(Joke.type_id == type_id).count()
+    # Check both old single-field and new array-field associations
+    old_count = db.query(Joke).filter(Joke.type_id == type_id).count()
+    all_jokes = db.query(Joke.type_ids).filter(Joke.type_ids.isnot(None)).all()
+    new_count = sum(1 for (raw,) in all_jokes if _parse_ids(raw) and type_id in _parse_ids(raw))
+    joke_count = old_count + new_count
     if joke_count > 0:
         raise HTTPException(status_code=400, detail=f"有 {joke_count} 条笑话使用此类型，无法删除")
     db.delete(db_type)
@@ -263,10 +340,165 @@ def admin_delete_crowd(
     db_crowd = db.query(JokeCrowd).filter(JokeCrowd.id == crowd_id).first()
     if not db_crowd:
         raise HTTPException(status_code=404, detail="人群分类不存在")
-    # Check if there are jokes using this crowd
-    joke_count = db.query(Joke).filter(Joke.crowd_id == crowd_id).count()
+    # Check both old single-field and new array-field associations
+    old_count = db.query(Joke).filter(Joke.crowd_id == crowd_id).count()
+    all_jokes = db.query(Joke.crowd_ids).filter(Joke.crowd_ids.isnot(None)).all()
+    new_count = sum(1 for (raw,) in all_jokes if _parse_ids(raw) and crowd_id in _parse_ids(raw))
+    joke_count = old_count + new_count
     if joke_count > 0:
         raise HTTPException(status_code=400, detail=f"有 {joke_count} 条笑话使用此人群，无法删除")
     db.delete(db_crowd)
     db.commit()
+    return {"message": "删除成功"}
+
+
+# ============================================================
+# 友情链接管理
+# ============================================================
+
+@router.get("/links", response_model=list[LinkResponse])
+def admin_list_links(
+    db: Session = Depends(get_db),
+    current_user: AdminUser = Depends(get_current_admin_user),
+):
+    """获取所有友情链接"""
+    return db.query(Link).order_by(Link.sort_order, Link.id).all()
+
+
+@router.post("/links", response_model=LinkResponse)
+def admin_create_link(
+    link: LinkCreate,
+    db: Session = Depends(get_db),
+    current_user: AdminUser = Depends(get_current_admin_user),
+):
+    """创建友情链接"""
+    db_link = Link(**link.model_dump())
+    db.add(db_link)
+    db.commit()
+    db.refresh(db_link)
+    return db_link
+
+
+@router.put("/links/{link_id}", response_model=LinkResponse)
+def admin_update_link(
+    link_id: int,
+    link: LinkCreate,
+    db: Session = Depends(get_db),
+    current_user: AdminUser = Depends(get_current_admin_user),
+):
+    """更新友情链接"""
+    db_link = db.query(Link).filter(Link.id == link_id).first()
+    if not db_link:
+        raise HTTPException(status_code=404, detail="链接不存在")
+    for key, value in link.model_dump().items():
+        setattr(db_link, key, value)
+    db.commit()
+    db.refresh(db_link)
+    return db_link
+
+
+@router.delete("/links/{link_id}")
+def admin_delete_link(
+    link_id: int,
+    db: Session = Depends(get_db),
+    current_user: AdminUser = Depends(get_current_admin_user),
+):
+    """删除友情链接"""
+    db_link = db.query(Link).filter(Link.id == link_id).first()
+    if not db_link:
+        raise HTTPException(status_code=404, detail="链接不存在")
+    db.delete(db_link)
+    db.commit()
+    return {"message": "删除成功"}
+
+
+# ============================================================
+# 反馈建议管理
+# ============================================================
+
+@router.get("/feedbacks", response_model=list[FeedbackResponse])
+def admin_list_feedbacks(
+    db: Session = Depends(get_db),
+    current_user: AdminUser = Depends(get_current_admin_user),
+):
+    """获取所有反馈建议"""
+    return db.query(Feedback).order_by(Feedback.created_at.desc()).all()
+
+
+@router.delete("/feedbacks/{feedback_id}")
+def admin_delete_feedback(
+    feedback_id: int,
+    db: Session = Depends(get_db),
+    current_user: AdminUser = Depends(get_current_admin_user),
+):
+    """删除反馈"""
+    db_feedback = db.query(Feedback).filter(Feedback.id == feedback_id).first()
+    if not db_feedback:
+        raise HTTPException(status_code=404, detail="反馈不存在")
+    db.delete(db_feedback)
+    db.commit()
+    return {"message": "删除成功"}
+
+
+# ============================================================
+# 用户管理
+# ============================================================
+
+@router.get("/users", response_model=list[UserResponse])
+def admin_list_users(
+    db: Session = Depends(get_db),
+    current_user: AdminUser = Depends(get_current_admin_user),
+):
+    """获取所有用户"""
+    if current_user.role != "admin":
+        raise HTTPException(status_code=403, detail="仅管理员可操作")
+    return db.query(AdminUser).order_by(AdminUser.created_at.desc()).all()
+
+
+@router.post("/users", response_model=UserResponse)
+def admin_create_user(
+    user: UserCreate,
+    db: Session = Depends(get_db),
+    current_user: AdminUser = Depends(get_current_admin_user),
+):
+    """创建用户"""
+    if current_user.role != "admin":
+        raise HTTPException(status_code=403, detail="仅管理员可操作")
+
+    # 检查用户名是否已存在
+    existing = db.query(AdminUser).filter(AdminUser.username == user.username).first()
+    if existing:
+        raise HTTPException(status_code=400, detail="用户名已存在")
+
+    db_user = AdminUser(
+        username=user.username,
+        password_hash=hash_password(user.password),
+        role=user.role,
+    )
+    db.add(db_user)
+    db.commit()
+    db.refresh(db_user)
+    return db_user
+
+
+@router.delete("/users/{user_id}")
+def admin_delete_user(
+    user_id: int,
+    db: Session = Depends(get_db),
+    current_user: AdminUser = Depends(get_current_admin_user),
+):
+    """删除用户"""
+    if current_user.role != "admin":
+        raise HTTPException(status_code=403, detail="仅管理员可操作")
+
+    db_user = db.query(AdminUser).filter(AdminUser.id == user_id).first()
+    if not db_user:
+        raise HTTPException(status_code=404, detail="用户不存在")
+
+    # 不允许删除自己
+    if db_user.id == current_user.id:
+        raise HTTPException(status_code=400, detail="不能删除自己")
+
+    db.delete(db_user)
+    db.commit()
     return {"message": "删除成功"}

api/app/routers/auth.py

@@ -9,10 +9,11 @@ from app.config import JWT_ALGORITHM, JWT_EXPIRATION_HOURS, JWT_SECRET_KEY
 from app.database import get_db
 from app.models.user import AdminUser
 from app.schemas.auth import LoginRequest, TokenResponse
+from app.schemas.user import UserCreate, UserResponse, UserRegister
 
 router = APIRouter(prefix="/auth", tags=["认证"])
 
-
+# 已存在的函数保持不变...
 def verify_password(plain_password: str, hashed_password: str) -> bool:
     """验证密码"""
     return bcrypt.checkpw(plain_password.encode(), hashed_password.encode())
@@ -27,11 +28,21 @@ def create_access_token(data: dict) -> str:
     return encoded_jwt
 
 
+def hash_password(password: str) -> str:
+    """哈希密码"""
+    return bcrypt.hashpw(password.encode(), bcrypt.gensalt()).decode()
+
+
+def get_current_user():
+    """获取当前用户（保留用于后续权限控制）"""
+    pass
+
+
 @router.post("/login", response_model=TokenResponse)
 def login(req: LoginRequest, db: Session = Depends(get_db)):
-    """管理员登录"""
+    """登录"""
     user = db.query(AdminUser).filter(AdminUser.username == req.username).first()
     if not user or not verify_password(req.password, user.password_hash):
         raise HTTPException(status_code=401, detail="用户名或密码错误")
-    access_token = create_access_token(data={"sub": str(user.id), "username": user.username})
+    access_token = create_access_token(data={"sub": str(user.id), "username": user.username, "role": user.role})
     return TokenResponse(access_token=access_token)

api/app/routers/feedback.py

@@ -0,0 +1,18 @@
+from fastapi import APIRouter, Depends
+from sqlalchemy.orm import Session
+
+from app.database import get_db
+from app.models.feedback import Feedback
+from app.schemas.feedback import FeedbackCreate, FeedbackResponse
+
+router = APIRouter(tags=["反馈建议"])
+
+
+@router.post("/feedback", response_model=FeedbackResponse)
+def create_feedback(feedback: FeedbackCreate, db: Session = Depends(get_db)):
+    """公开：提交反馈建议"""
+    db_feedback = Feedback(**feedback.model_dump())
+    db.add(db_feedback)
+    db.commit()
+    db.refresh(db_feedback)
+    return db_feedback

api/app/routers/generate.py

@@ -0,0 +1,187 @@
+"""AI 笑话生成器 API — 提示词从数据库读取"""
+
+import json
+from datetime import datetime
+from fastapi import APIRouter, Depends, HTTPException
+from openai import OpenAI
+from sqlalchemy.orm import Session
+
+from app.database import get_db
+from app.models.setting import AiSetting
+from app.schemas.joke import GenerateRequest, GenerateResponse
+
+
+router = APIRouter(prefix="/generate", tags=["生成器"])
+
+STYLE_MAP = {
+    "cold": "冷幽默 / 无厘头",
+    "warm": "温馨幽默 / 暖心搞笑",
+    "twist": "反转 / 神转折",
+    "pun": "谐音梗 / 文字游戏",
+    "sketch": "段子 / 吐槽调侃",
+    "irony": "讽刺幽默 / 黑色幽默",
+}
+
+LENGTH_MAP = {
+    "short": "30-80字，非常简短",
+    "medium": "80-150字，正常长度",
+    "long": "150-300字，可以描述一个小场景",
+}
+
+
+def _load_prompt(setting: AiSetting) -> str:
+    """从数据库读取生成提示词，没有则返回默认值"""
+    prompt = setting.generate_prompt
+    if not prompt or not prompt.strip():
+        prompt = GENERATE_PROMPT_DEFAULT
+    return prompt
+
+
+GENERATE_PROMPT_DEFAULT = """你是一位幽默大师，专门创作轻松搞笑的短笑话。
+
+{context}
+
+要求：
+1. 根据场景和关键词创作一条原创笑话
+2. 笑话要有反转或意外结局
+3. 语言风格：{style}
+4. 字数要求：{length_requirement}
+5. 直接输出笑话内容，不需要解释
+
+请严格按照以下 JSON 格式输出，不要加任何额外说明：
+{{
+  "title": "笑话标题",
+  "content": "笑话正文",
+  "score": 8,
+  "reason": "这个笑话巧妙结合了场景和关键词，结尾有反转"
+}}
+
+score 是 1-10 的整数评分，reason 是用一句话说明亮点。"""
+
+
+def _build_prompt(
+    scenarios: list[str],
+    keywords: list[str],
+    style: str = "twist",
+    length: str = "medium",
+    setting: AiSetting | None = None,
+) -> str:
+    """构建 AI prompt，支持风格和长度控制"""
+    parts = []
+    if scenarios:
+        parts.append(f"场景：{', '.join(scenarios)}")
+    if keywords:
+        parts.append(f"关键词：{', '.join(keywords)}")
+    if not parts:
+        parts.append("场景：日常生活的各种趣事（不指定具体场景）")
+
+    prompt_template = _load_prompt(setting) if setting else GENERATE_PROMPT_DEFAULT
+    return prompt_template.format(
+        context="\n".join(parts),
+        style=STYLE_MAP.get(style, "反转 / 神转折"),
+        length_requirement=LENGTH_MAP.get(length, "80-150字，正常长度"),
+    )
+
+
+@router.post("", response_model=GenerateResponse)
+def generate_joke(
+    req: GenerateRequest,
+    db: Session = Depends(get_db),
+):
+    """调用 AI 生成笑话，提示词从数据库读取"""
+    # 获取激活的 AI 配置
+    setting = db.query(AiSetting).filter(AiSetting.is_active == True).first()
+    if not setting:
+        setting = db.query(AiSetting).first()
+
+    if not setting or not setting.api_key:
+        raise HTTPException(status_code=503, detail="AI 服务未配置，请联系管理员")
+
+    try:
+        client = OpenAI(base_url=setting.api_base, api_key=setting.api_key)
+        prompt = _build_prompt(req.scenarios, req.keywords, req.style, req.length, setting)
+
+        response = client.chat.completions.create(
+            model=setting.model_name,
+            messages=[{"role": "user", "content": prompt}],
+            temperature=req.temperature,
+            max_tokens=setting.max_tokens,
+        )
+
+        raw = response.choices[0].message.content
+        return _parse_json_response(raw)
+
+    except json.JSONDecodeError:
+        raise HTTPException(status_code=500, detail="AI 返回格式异常，请重新生成")
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=f"生成失败: {str(e)}")
+
+
+def _parse_json_response(raw: str | None) -> GenerateResponse:
+    """解析 AI 返回的 JSON 格式"""
+    if not raw or not raw.strip():
+        return GenerateResponse(title="生成的笑话", content="（内容生成失败，请重新生成）", score=0)
+
+    raw = raw.strip()
+    if raw.startswith("```"):
+        lines = raw.split("\n")
+        if len(lines) >= 3:
+            raw = "\n".join(lines[1:-1]).strip()
+
+    if raw.endswith(","):
+        raw = raw[:-1]
+
+    try:
+        data = json.loads(raw)
+    except json.JSONDecodeError:
+        try:
+            start = raw.index("{")
+            end = raw.rindex("}") + 1
+            raw = raw[start:end]
+            data = json.loads(raw)
+        except (ValueError, json.JSONDecodeError):
+            return _parse_response(raw)
+
+    return GenerateResponse(
+        title=data.get("title", "生成的笑话"),
+        content=data.get("content", ""),
+        score=data.get("score", 0),
+        reason=data.get("reason", ""),
+        created_at=datetime.now(),
+    )
+
+
+def _parse_response(raw: str | None) -> GenerateResponse:
+    """解析 AI 返回内容（非 JSON 回退）"""
+    if not raw or not raw.strip():
+        raise ValueError("AI 返回内容为空")
+
+    title = ""
+    content = raw
+
+    for line in raw.split("\n"):
+        line = line.strip()
+        if line.startswith("标题：") or line.startswith("标题:"):
+            title = line.split("：", 1)[-1].split(":", 1)[-1].strip()
+            lines = content.split("\n")
+            for i, l in enumerate(lines):
+                if l.strip() == line:
+                    lines[i] = ""
+                    break
+            content = "\n".join(lines).strip()
+            break
+
+    if not title:
+        first_line = raw.split("\n")[0].strip()
+        if first_line.startswith("标题"):
+            first_line = first_line.split("：", 1)[-1].split(":", 1)[-1].strip()
+        title = first_line[:30] if len(first_line) > 30 else first_line
+
+    if not content.strip():
+        content = "（内容生成失败，请重新生成）"
+
+    return GenerateResponse(
+        title=title or "生成的笑话",
+        content=content.strip(),
+        created_at=datetime.now(),
+    )

api/app/routers/jokes.py

@@ -1,78 +1,285 @@
+import json
 import random
+from datetime import datetime
 
 from fastapi import APIRouter, Depends, HTTPException, Query
-from sqlalchemy import func
+from sqlalchemy import func, text
 from sqlalchemy.orm import Session
 
 from app.database import get_db
 from app.models.joke import Joke
+from app.models.category import JokeCrowd, JokeType
 from app.schemas.joke import JokeResponse, PaginatedJokeResponse
 
 router = APIRouter(prefix="/jokes", tags=["笑话"])
 
 
-def joke_to_response(joke: Joke) -> JokeResponse:
-    """Convert Joke model to JokeResponse schema."""
+def _parse_ids(raw) -> list[int]:
+    if not raw:
+        return []
+    if isinstance(raw, list):
+        return raw
+    try:
+        return json.loads(raw)
+    except Exception:
+        return []
+
+
+def _batch_load_categories(db: Session, jokes: list[Joke]) -> tuple[dict, dict]:
+    """批量加载类型和人群名称，避免 N+1 查询"""
+    # 收集所有需要的 id
+    all_type_ids = set()
+    all_crowd_ids = set()
+    for joke in jokes:
+        all_type_ids.update(_parse_ids(joke.type_ids))
+        all_crowd_ids.update(_parse_ids(joke.crowd_ids))
+
+    # 批量查询
+    type_map = {}
+    if all_type_ids:
+        types = db.query(JokeType).filter(JokeType.id.in_(all_type_ids)).all()
+        type_map = {t.id: t.name for t in types}
+
+    crowd_map = {}
+    if all_crowd_ids:
+        crowds = db.query(JokeCrowd).filter(JokeCrowd.id.in_(all_crowd_ids)).all()
+        crowd_map = {c.id: c.name for c in crowds}
+
+    return type_map, crowd_map
+
+
+def joke_to_response(joke: Joke, type_map: dict = None, crowd_map: dict = None) -> JokeResponse:
+    """Convert Joke model to JokeResponse schema.
+
+    Args:
+        type_map: 预加载的类型 id→name 映射
+        crowd_map: 预加载的人群 id→name 映射
+    """
+    ids = _parse_ids(joke.type_ids)
+    crowd_ids = _parse_ids(joke.crowd_ids)
+
+    if type_map is not None:
+        type_names = [type_map[i] for i in ids if i in type_map]
+    else:
+        type_names = []
+
+    if crowd_map is not None:
+        crowd_names = [crowd_map[i] for i in crowd_ids if i in crowd_map]
+    else:
+        crowd_names = []
+
     return JokeResponse(
         id=joke.id,
         title=joke.title,
         content=joke.content,
-        type_id=joke.type_id,
-        crowd_id=joke.crowd_id,
+        type_ids=ids,
+        crowd_ids=crowd_ids,
         status=joke.status,
         view_count=joke.view_count,
         like_count=joke.like_count,
+        dislike_count=joke.dislike_count,
         created_at=joke.created_at,
         updated_at=joke.updated_at,
-        type_name=joke.type.name if joke.type else None,
-        crowd_name=joke.crowd.name if joke.crowd else None,
+        type_names=type_names,
+        crowd_names=crowd_names,
+        polished_content=joke.polished_content,
+        ai_score=joke.ai_score,
+        ai_level=joke.ai_level,
     )
 
 
+def _build_json_filter(column, target_ids: set[int]) -> list:
+    """构建精确的 JSON 数组匹配过滤器（SQLite）"""
+    filters = []
+    for tid in target_ids:
+        # 匹配 [n,...] 或 [...,n,...] 或 [...,n]
+        # 使用 JSON_EACH 函数检查
+        json_str = json.dumps(tid)
+        filters.append(
+            func.json_extract(column, '$').cast(type(target_ids)).contains(tid)
+        )
+    return filters
+
+
 @router.get("/", response_model=PaginatedJokeResponse)
 def list_jokes(
     page: int = 1,
     page_size: int = 20,
-    type_id: int | None = None,
-    crowd_id: int | None = None,
+    type_ids: str | None = Query(None, description="逗号分隔的类型 ID，如 1,3,5"),
+    crowd_ids: str | None = Query(None, description="逗号分隔的人群 ID，如 2,4"),
+    keyword: str | None = Query(None, description="关键词搜索（标题+内容）"),
     db: Session = Depends(get_db),
 ):
     """获取笑话列表（仅返回已审核通过的笑话）"""
     query = db.query(Joke).filter(Joke.status == "approved")
 
-    if type_id is not None:
-        query = query.filter(Joke.type_id == type_id)
-    if crowd_id is not None:
-        query = query.filter(Joke.crowd_id == crowd_id)
+    # 关键词搜索：标题和内容模糊匹配
+    if keyword:
+        kw = f"%{keyword}%"
+        query = query.filter(
+            (Joke.title.like(kw)) | (Joke.content.like(kw))
+        )
+
+    # 支持多选过滤：逗号分隔的 ID
+    # 使用自定义 JSON 匹配函数避免 LIKE %N% 的错误匹配
+    if type_ids:
+        filter_set = {int(x.strip()) for x in type_ids.split(",") if x.strip().isdigit()}
+        if filter_set:
+            # 使用文本匹配，确保精确匹配 JSON 数组中的 id
+            # 匹配模式：[1,2,3] 或 [1] 或带尾部逗号的
+            type_filters = []
+            for tid in filter_set:
+                # 构建精确匹配的正则：在逗号、引号、方括号包围中的数字
+                import re
+                pattern = rf'["\s\[,]{tid}[",\s\]]'
+                type_filters.append(Joke.type_ids.regexp_match(pattern))
+            # 只要匹配任一类型即可
+            type_filter = type_filters[0]
+            for f in type_filters[1:]:
+                type_filter = type_filter | f
+            # 还需要兼容旧的单值 type_id 字段
+            old_filter = Joke.type_id.in_(filter_set)
+            query = query.filter(old_filter | type_filter)
+
+    if crowd_ids:
+        filter_set = {int(x.strip()) for x in crowd_ids.split(",") if x.strip().isdigit()}
+        if filter_set:
+            import re
+            crowd_filters = []
+            for cid in filter_set:
+                pattern = rf'["\s\[,]{cid}[",\s\]]'
+                crowd_filters.append(Joke.crowd_ids.regexp_match(pattern))
+            crowd_filter = crowd_filters[0]
+            for f in crowd_filters[1:]:
+                crowd_filter = crowd_filter | f
+            old_filter = Joke.crowd_id.in_(filter_set)
+            query = query.filter(old_filter | crowd_filter)
 
     total = query.count()
     offset = (page - 1) * page_size
     jokes = query.order_by(Joke.created_at.desc()).offset(offset).limit(page_size).all()
 
+    # 批量预加载类型和人群名称
+    type_map, crowd_map = _batch_load_categories(db, jokes)
+
     return PaginatedJokeResponse(
-        items=[joke_to_response(j) for j in jokes],
+        items=[joke_to_response(j, type_map, crowd_map) for j in jokes],
         total=total,
         page=page,
         page_size=page_size,
     )
 
 
-@router.get("/{joke_id}", response_model=JokeResponse)
-def get_joke(joke_id: int, db: Session = Depends(get_db)):
-    """获取单条笑话详情"""
-    joke = db.query(Joke).filter(Joke.id == joke_id).first()
-    if not joke:
-        raise HTTPException(status_code=404, detail="笑话不存在")
-    # 增加浏览次数
-    joke.view_count += 1
-    db.commit()
-    return joke_to_response(joke)
-
-
 @router.get("/random", response_model=JokeResponse)
 def get_random_joke(db: Session = Depends(get_db)):
     """随机获取一条已审核通过的笑话"""
+    # 使用效率更高的方式：随机 ID 取模
+    max_id = db.query(func.max(Joke.id)).filter(Joke.status == "approved").scalar()
+    if not max_id:
+        raise HTTPException(status_code=404, detail="暂无笑话")
+
+    # 尝试最多 10 次找到有效笑话
+    for _ in range(10):
+        random_id = random.randint(1, max_id)
+        joke = db.query(Joke).filter(
+            Joke.id >= random_id,
+            Joke.status == "approved"
+        ).first()
+        if joke:
+            type_map, crowd_map = _batch_load_categories(db, [joke])
+            return joke_to_response(joke, type_map, crowd_map)
+
+    # 兜底：全表随机
     joke = db.query(Joke).filter(Joke.status == "approved").order_by(func.random()).first()
     if not joke:
         raise HTTPException(status_code=404, detail="暂无笑话")
-    return joke_to_response(joke)
+    type_map, crowd_map = _batch_load_categories(db, [joke])
+    return joke_to_response(joke, type_map, crowd_map)
+
+
+@router.get("/hot-monthly")
+def hot_monthly_jokes(db: Session = Depends(get_db)):
+    """当月热门笑话：本月浏览量最高的前 10 条"""
+    now = datetime.now()
+    start_of_month = now.replace(day=1, hour=0, minute=0, second=0, microsecond=0)
+    jokes = (
+        db.query(Joke)
+        .filter(
+            Joke.status == "approved",
+            Joke.created_at >= start_of_month,
+        )
+        .order_by(Joke.view_count.desc())
+        .limit(10)
+        .all()
+    )
+    type_map, crowd_map = _batch_load_categories(db, jokes)
+    return [joke_to_response(j, type_map, crowd_map) for j in jokes]
+
+
+@router.get("/stats")
+def public_stats(db: Session = Depends(get_db)):
+    """公开统计数据：笑话总数、审核/待审/拒绝数量、总浏览/赞/踩"""
+    total = db.query(Joke).count()
+    approved = db.query(Joke).filter(Joke.status == "approved").count()
+    pending = db.query(Joke).filter(Joke.status == "pending").count()
+    rejected = db.query(Joke).filter(Joke.status == "rejected").count()
+    total_views = db.query(Joke).with_entities(func.sum(Joke.view_count)).scalar() or 0
+    total_likes = db.query(Joke).with_entities(func.sum(Joke.like_count)).scalar() or 0
+    total_dislikes = db.query(Joke).with_entities(func.sum(Joke.dislike_count)).scalar() or 0
+    return {
+        "total_jokes": total,
+        "approved_jokes": approved,
+        "pending_jokes": pending,
+        "rejected_jokes": rejected,
+        "total_views": total_views,
+        "total_likes": total_likes,
+        "total_dislikes": total_dislikes,
+    }
+
+
+@router.get("/{joke_id}", response_model=JokeResponse)
+def get_joke(joke_id: int, db: Session = Depends(get_db)):
+    """获取单条笑话详情"""
+    # 只返回已审核通过的笑话
+    joke = db.query(Joke).filter(
+        Joke.id == joke_id,
+        Joke.status == "approved"
+    ).first()
+    if not joke:
+        raise HTTPException(status_code=404, detail="笑话不存在")
+    # 使用原子更新避免并发竞态
+    db.query(Joke).filter(Joke.id == joke_id).update({Joke.view_count: Joke.view_count + 1})
+    db.commit()
+    # 重新查询获取更新后的数据
+    db.refresh(joke)
+    # 批量加载类型和人群名称
+    type_map, crowd_map = _batch_load_categories(db, [joke])
+    return joke_to_response(joke, type_map, crowd_map)
+
+
+def _vote_joke(joke_id: int, field: str, db: Session):
+    """通用投票逻辑：点赞或点踩（仅允许已审核通过的笑话）"""
+    joke = db.query(Joke).filter(
+        Joke.id == joke_id,
+        Joke.status == "approved"
+    ).first()
+    if not joke:
+        raise HTTPException(status_code=404, detail="笑话不存在")
+    # 使用原子更新避免并发竞态
+    update_data = {getattr(Joke, field): getattr(Joke, field) + 1}
+    db.query(Joke).filter(Joke.id == joke_id).update(update_data)
+    db.commit()
+    db.refresh(joke)
+    return {"message": "操作成功", field: getattr(joke, field)}
+
+
+@router.post("/{joke_id}/like")
+def like_joke(joke_id: int, db: Session = Depends(get_db)):
+    """为笑话点赞（仅允许已审核通过的笑话）"""
+    return _vote_joke(joke_id, "like_count", db)
+
+
+@router.post("/{joke_id}/dislike")
+def dislike_joke(joke_id: int, db: Session = Depends(get_db)):
+    """为笑话点踩（仅允许已审核通过的笑话）"""
+    return _vote_joke(joke_id, "dislike_count", db)

api/app/routers/links.py

@@ -0,0 +1,38 @@
+from fastapi import APIRouter, Depends
+from sqlalchemy.orm import Session
+
+from app.database import get_db
+from app.models.link import Link
+from app.schemas.link import LinkCreate, LinkResponse, LinkApply
+
+router = APIRouter(tags=["友情链接"])
+
+
+@router.get("/links", response_model=list[LinkResponse])
+def list_links(db: Session = Depends(get_db)):
+    """公开：获取已通过审核的友情链接"""
+    return db.query(Link).filter(Link.status == "approved").order_by(Link.sort_order, Link.id).all()
+
+
+@router.post("/links/apply", response_model=LinkResponse)
+def apply_link(link_data: LinkApply, db: Session = Depends(get_db)):
+    """公开：申请友链"""
+    # 检查是否已存在同名或同URL的申请/已通过友链
+    existing = db.query(Link).filter(
+        (Link.name == link_data.name) | (Link.url == link_data.url)
+    ).first()
+    if existing:
+        return existing  # 已存在则返回已有记录
+
+    link = Link(
+        name=link_data.name,
+        url=link_data.url,
+        description=link_data.description,
+        contact=link_data.contact,
+        status="pending",
+        sort_order=999,  # 新申请放在最后
+    )
+    db.add(link)
+    db.commit()
+    db.refresh(link)
+    return link

api/app/routers/settings.py

@@ -0,0 +1,114 @@
+from fastapi import APIRouter, Depends, HTTPException
+from sqlalchemy.orm import Session
+
+from app.database import get_db
+from app.models.setting import AiSetting
+from app.schemas.setting import AiSettingCreate
+from app.routers.admin import get_current_admin_user
+from app.models.user import AdminUser
+
+router = APIRouter(prefix="/admin/settings", tags=["AI设置"])
+
+
+@router.get("", response_model=list)
+def list_settings(
+    db: Session = Depends(get_db),
+    current_user: AdminUser = Depends(get_current_admin_user),
+):
+    """列出所有 AI 配置"""
+    return db.query(AiSetting).order_by(AiSetting.created_at.desc()).all()
+
+
+@router.get("/active")
+def get_active_setting(
+    db: Session = Depends(get_db),
+):
+    """获取当前激活的 AI 配置（无需认证，供爬虫/优化器使用）"""
+    setting = db.query(AiSetting).filter(AiSetting.is_active == True).first()
+    if not setting:
+        from app.routers.settings import _ensure_default_settings
+        setting = _ensure_default_settings(db)
+    return setting
+
+
+@router.post("")
+def create_setting(
+    setting: AiSettingCreate,
+    db: Session = Depends(get_db),
+    current_user: AdminUser = Depends(get_current_admin_user),
+):
+    """新建 AI 配置"""
+    db_setting = AiSetting(**setting.model_dump())
+    db.add(db_setting)
+    db.commit()
+    db.refresh(db_setting)
+    return db_setting
+
+
+@router.put("/{setting_id}")
+def update_setting(
+    setting_id: int,
+    setting: AiSettingCreate,
+    db: Session = Depends(get_db),
+    current_user: AdminUser = Depends(get_current_admin_user),
+):
+    """更新 AI 配置"""
+    db_setting = db.query(AiSetting).filter(AiSetting.id == setting_id).first()
+    if not db_setting:
+        raise HTTPException(status_code=404, detail="配置不存在")
+    for key, value in setting.model_dump().items():
+        setattr(db_setting, key, value)
+    db.commit()
+    db.refresh(db_setting)
+    return db_setting
+
+
+@router.put("/{setting_id}/toggle")
+def toggle_setting(
+    setting_id: int,
+    db: Session = Depends(get_db),
+    current_user: AdminUser = Depends(get_current_admin_user),
+):
+    """切换激活状态（只能有一个活跃）"""
+    db_setting = db.query(AiSetting).filter(AiSetting.id == setting_id).first()
+    if not db_setting:
+        raise HTTPException(status_code=404, detail="配置不存在")
+
+    # 先全部设为非活跃
+    db.query(AiSetting).update({AiSetting.is_active: False})
+    db_setting.is_active = True
+    db.commit()
+    return {"message": "已激活", "id": setting_id}
+
+
+@router.delete("/{setting_id}")
+def delete_setting(
+    setting_id: int,
+    db: Session = Depends(get_db),
+    current_user: AdminUser = Depends(get_current_admin_user),
+):
+    """删除 AI 配置"""
+    db_setting = db.query(AiSetting).filter(AiSetting.id == setting_id).first()
+    if not db_setting:
+        raise HTTPException(status_code=404, detail="配置不存在")
+    db.delete(db_setting)
+    db.commit()
+    return {"message": "删除成功"}
+
+
+def _ensure_default_settings(db: Session) -> AiSetting:
+    """当没有激活配置时，创建一条默认配置"""
+    existing = db.query(AiSetting).first()
+    if existing:
+        existing.is_active = True
+        db.commit()
+        return existing
+
+    defaults = AiSetting(
+        is_active=True,
+        # 默认提示词会在 main.py 迁移时填充
+    )
+    db.add(defaults)
+    db.commit()
+    db.refresh(defaults)
+    return defaults

api/app/routers/submit.py

@@ -0,0 +1,62 @@
+"""
+公开提交流程 - 供外部用户（submitter）使用。
+无后台管理权限，仅能提交笑话到 pending 状态。
+"""
+from fastapi import APIRouter, Depends, HTTPException
+from sqlalchemy.orm import Session
+from pydantic import BaseModel
+
+from app.database import get_db
+from app.models.joke import Joke
+
+router = APIRouter(tags=["笑话提交"])
+
+
+class JokeSubmit(BaseModel):
+    """笑话提交"""
+    title: str
+    content: str
+
+
+class JokeBatchSubmit(BaseModel):
+    """批量提交"""
+    jokes: list[JokeSubmit]
+
+
+@router.post("/submit", response_model=dict)
+def submit_joke(data: JokeSubmit, db: Session = Depends(get_db)):
+    """公开接口：提交单条笑话（自动进入待审核状态）"""
+    joke = Joke(
+        title=data.title,
+        content=data.content,
+        status="pending",
+    )
+    db.add(joke)
+    db.commit()
+    db.refresh(joke)
+    return {
+        "success": True,
+        "id": joke.id,
+        "message": f"已提交，ID: {joke.id}"
+    }
+
+
+@router.post("/submit/batch", response_model=dict)
+def submit_jokes_batch(data: JokeBatchSubmit, db: Session = Depends(get_db)):
+    """公开接口：批量提交笑话"""
+    added = []
+    for item in data.jokes:
+        joke = Joke(
+            title=item.title,
+            content=item.content,
+            status="pending",
+        )
+        db.add(joke)
+        added.append(item.title)
+
+    db.commit()
+    return {
+        "success": True,
+        "count": len(added),
+        "message": f"已提交 {len(added)} 条笑话"
+    }

api/app/schemas/feedback.py

@@ -0,0 +1,17 @@
+from datetime import datetime
+
+from pydantic import BaseModel
+
+
+class FeedbackCreate(BaseModel):
+    name: str | None = None
+    email: str | None = None
+    content: str
+
+
+class FeedbackResponse(FeedbackCreate):
+    id: int
+    created_at: datetime
+
+    class Config:
+        from_attributes = True

api/app/schemas/joke.py

@@ -6,8 +6,8 @@ from pydantic import BaseModel
 class JokeBase(BaseModel):
     title: str
     content: str
-    type_id: int | None = None
-    crowd_id: int | None = None
+    type_ids: list[int] | None = None
+    crowd_ids: list[int] | None = None
 
 
 class JokeCreate(JokeBase):
@@ -17,27 +17,86 @@ class JokeCreate(JokeBase):
 class JokeUpdate(BaseModel):
     title: str | None = None
     content: str | None = None
-    type_id: int | None = None
-    crowd_id: int | None = None
+    polished_content: str | None = None
+    ai_score: float | None = None
+    ai_level: str | None = None
+    type_ids: list[int] | None = None
+    crowd_ids: list[int] | None = None
     status: str | None = None
 
 
 class JokeResponse(JokeBase):
     id: int
     status: str
+    polished_content: str | None = None
+    # AI 评价字段
+    ai_score: float | None = None
+    ai_level: str | None = None
     view_count: int
     like_count: int
+    dislike_count: int
     created_at: datetime
     updated_at: datetime
-    type_name: str | None = None
-    crowd_name: str | None = None
+    type_names: list[str] | None = None
+    crowd_names: list[str] | None = None
 
     class Config:
         from_attributes = True
 
+    @property
+    def display_content(self) -> str:
+        """优先显示润色后的内容"""
+        return self.polished_content if self.polished_content else self.content
+
+    def get_level_display(self) -> str:
+        """获取 AI 评级的中文显示"""
+        level_map = {
+            'excellent': '⭐⭐⭐ 精品',
+            'good': '⭐⭐ 良好',
+            'ordinary': '⭐ 普通',
+            'poor': '⚠️ 待优化'
+        }
+        return level_map.get(self.ai_level, '') if self.ai_level else ''
+
 
 class PaginatedJokeResponse(BaseModel):
     items: list[JokeResponse]
     total: int
     page: int
-    page_size: int
+    page_size: int
+
+
+# ============================================================
+# 笑话生成器 Schema
+# ============================================================
+
+class GenerateRequest(BaseModel):
+    """笑话生成请求"""
+    keywords: list[str] = []
+    scenarios: list[str] = []
+    style: str = "twist"
+    length: str = "medium"
+    temperature: float = 0.8
+
+    class Config:
+        json_schema_extra = {
+            "example": {
+                "keywords": ["加班"],
+                "scenarios": ["职场"],
+                "style": "twist",
+                "length": "medium",
+                "temperature": 0.8,
+            }
+        }
+
+
+class GenerateResponse(BaseModel):
+    """笑话生成响应"""
+    title: str
+    content: str
+    score: int = 0
+    reason: str = ""
+    created_at: datetime | None = None
+
+    class Config:
+        from_attributes = True

api/app/schemas/link.py

@@ -0,0 +1,28 @@
+from datetime import datetime
+
+from pydantic import BaseModel, HttpUrl
+
+
+class LinkCreate(BaseModel):
+    name: str
+    url: str
+    description: str | None = None
+    sort_order: int = 0
+
+
+class LinkApply(BaseModel):
+    """友链申请"""
+    name: str
+    url: str
+    description: str | None = None
+    contact: str | None = None  # 联系方式
+
+
+class LinkResponse(LinkCreate):
+    id: int
+    status: str
+    contact: str | None = None
+    created_at: datetime
+
+    class Config:
+        from_attributes = True

api/app/schemas/setting.py

@@ -0,0 +1,31 @@
+from pydantic import BaseModel, Field
+
+
+class AiSettingBase(BaseModel):
+    provider: str = Field(default="nvidia")
+    api_base: str = Field(default="https://integrate.api.nvidia.com/v1")
+    api_key: str = Field(default="")
+    model_name: str = Field(default="nvidia/llama-3.1-nemotron-70b-instruct")
+    temperature: float = Field(default=0.7, ge=0, le=2)
+    max_tokens: int = Field(default=2048, ge=1)
+
+    # 提示词
+    generate_prompt: str = Field(default="")
+    optimizer_quality_prompt: str = Field(default="")
+    optimizer_polish_prompt: str = Field(default="")
+    optimizer_evaluate_prompt: str = Field(default="")
+    crawler_extract_prompt: str = Field(default="")
+    crawler_rewrite_prompt: str = Field(default="")
+
+    # 优化器各阶段温度
+    optimizer_quality_temperature: float = Field(default=0.3, ge=0, le=2)
+    optimizer_polish_temperature: float = Field(default=0.8, ge=0, le=2)
+    optimizer_evaluate_temperature: float = Field(default=0.3, ge=0, le=2)
+
+    crawl_enabled: bool = Field(default=False)
+    crawl_keywords: str = Field(default="")
+    max_pages_per_run: int = Field(default=3, ge=1)
+
+
+class AiSettingCreate(AiSettingBase):
+    pass

api/app/schemas/user.py

@@ -0,0 +1,26 @@
+from datetime import datetime
+from pydantic import BaseModel
+
+
+class UserCreate(BaseModel):
+    """创建用户"""
+    username: str
+    password: str
+    role: str = "submitter"  # admin=管理员, submitter=提交者
+
+
+class UserResponse(BaseModel):
+    """用户信息"""
+    id: int
+    username: str
+    role: str
+    created_at: datetime
+
+    class Config:
+        from_attributes = True
+
+
+class UserRegister(BaseModel):
+    """公开注册（仅限提交者）"""
+    username: str
+    password: str

api/main.py

@@ -2,7 +2,130 @@ from fastapi import FastAPI
 from fastapi.middleware.cors import CORSMiddleware
 
 from app.config import API_TITLE, API_VERSION
-from app.routers import jokes_router, categories_router, auth_router, admin_router
+from app.routers import jokes_router, categories_router, auth_router, admin_router, settings_router, links_router, feedback_router, generate_router
+from app.routers.submit import router as submit_router
+from app.database import Base, engine
+from app.models.setting import AiSetting
+from app.models.link import Link
+from app.models.feedback import Feedback  # 注册模型，确保 create_all 能看到
+
+# 创建新表（如果不存在）
+Base.metadata.create_all(bind=engine)
+
+# 对已有表新增字段的兼容迁移（SQLite 不支持 ALTER TABLE ADD COLUMN IF NOT EXISTS）
+# 默认提示词
+_GENERATE_PROMPT_DEFAULT = """你是一位幽默大师，专门创作轻松搞笑的短笑话。
+
+{context}
+
+要求：
+1. 根据场景和关键词创作一条原创笑话
+2. 笑话要有反转或意外结局
+3. 语言风格：{style}
+4. 字数要求：{length_requirement}
+5. 直接输出笑话内容，不需要解释
+
+请严格按照以下 JSON 格式输出，不要加任何额外说明：
+{{
+  "title": "笑话标题",
+  "content": "笑话正文",
+  "score": 8,
+  "reason": "这个笑话巧妙结合了场景和关键词，结尾有反转"
+}}
+
+score 是 1-10 的整数评分，reason 是用一句话说明亮点。"""
+
+_OPTIMIZER_QUALITY_DEFAULT = """你是一个幽默内容审核专家。判断以下内容是否是一个合格的笑话/段子。
+
+合格标准（满足任一即可）：
+1. 有明确的笑点或反转（punchline）
+2. 有幽默的语言表达或双关
+3. 有意外结局或情理之中意料之外
+
+不合格标准（符合任一即判定不合格）：
+1. 纯粹的事实陈述，没有任何幽默元素
+2. 只是对话片段，没有笑点
+3. 普通故事或叙事，没有幽默设计
+4. 说教或道理阐述
+5. 内容不完整或难以理解
+
+始终返回 JSON 格式：{"has_punchline": true/false, "reason": "简要说明判断理由"}"""
+
+_OPTIMIZER_POLISH_DEFAULT = """你是一个专业的幽默文案编辑。请润色以下笑话，要求：
+1. 保持核心笑点不变
+2. 优化语言表达，使其更通顺、更精炼
+3. 增强节奏感和幽默效果，但不改变原意
+4. 字数控制在原内容的 80%-120%
+5. 不要添加额外解释或评论
+6. 直接输出润色后的内容，不要加任何前缀"""
+
+_OPTIMIZER_EVALUATE_DEFAULT = """你是一个笑话分类和评价专家。对给定的笑话进行分析，返回 JSON 格式的分类和评分结果。
+
+要求：
+1. types: 从提供的类型列表中选择所有匹配的类型名称（数组，可以选多个）
+2. crowds: 从提供的人群列表中选择所有匹配的人群名称（数组，可以选多个）
+3. score: 1-10 分，基于幽默程度、创意和表达效果
+4. comment: 简短评语（10字以内）
+
+始终返回 JSON 格式。"""
+
+_CRAWLER_EXTRACT_DEFAULT = """你是一个笑话提取专家。从给定的网页文本中识别并提取所有笑话、幽默段子或有趣内容。
+要求：
+1. 只返回真正的笑话内容，不要提取普通文章或新闻
+2. 每条笑话需要包含：title（简短标题）、content（完整笑话内容）、type（类型）、crowd（人群）
+3. 如果网页中没有笑话，返回空数组 []
+4. 永远返回合法的 JSON 格式，根节点为数组或包含 jokes 键的对象"""
+
+_CRAWLER_REWRITE_DEFAULT = """你是一个幽默作家，负责润色和改写笑话。
+要求：
+1. 保持笑话的核心笑点不变
+2. 语言更通顺、更幽默
+3. 字数控制在原内容的 80%-120% 之间
+4. 不要添加任何解释说明"""
+
+
+def _migrate_db():
+    from sqlalchemy import inspect, text
+    inspector = inspect(engine)
+    columns = [c["name"] for c in inspector.get_columns("ai_settings")]
+
+    # 新增提示词字段
+    prompt_fields = {
+        "generate_prompt": _GENERATE_PROMPT_DEFAULT,
+        "optimizer_quality_prompt": _OPTIMIZER_QUALITY_DEFAULT,
+        "optimizer_polish_prompt": _OPTIMIZER_POLISH_DEFAULT,
+        "optimizer_evaluate_prompt": _OPTIMIZER_EVALUATE_DEFAULT,
+        "crawler_extract_prompt": _CRAWLER_EXTRACT_DEFAULT,
+        "crawler_rewrite_prompt": _CRAWLER_REWRITE_DEFAULT,
+    }
+    for field, default_val in prompt_fields.items():
+        if field not in columns:
+            from sqlalchemy import Text
+            col_type = "TEXT"
+            with engine.connect() as conn:
+                conn.execute(text(f"ALTER TABLE ai_settings ADD COLUMN {field} {col_type}"))
+                conn.commit()
+
+    # 新增 temperature 字段
+    temp_fields = [
+        ("optimizer_quality_temperature", "FLOAT", "0.3"),
+        ("optimizer_polish_temperature", "FLOAT", "0.8"),
+        ("optimizer_evaluate_temperature", "FLOAT", "0.3"),
+    ]
+    for field, col_type, default_val in temp_fields:
+        if field not in columns:
+            with engine.connect() as conn:
+                conn.execute(text(f"ALTER TABLE ai_settings ADD COLUMN {field} {col_type} DEFAULT {default_val}"))
+                conn.commit()
+
+    # 给新增字段填充默认值（已有记录）
+    with engine.connect() as conn:
+        for field, default_val in prompt_fields.items():
+            conn.execute(text(f"UPDATE ai_settings SET {field} = :val WHERE {field} IS NULL"),
+                         {"val": default_val})
+        conn.commit()
+
+_migrate_db()
 
 # Create FastAPI application
 app = FastAPI(title=API_TITLE, version=API_VERSION)
@@ -17,10 +140,15 @@ app.add_middleware(
 )
 
 # Mount routers
-app.include_router(jokes_router)
-app.include_router(categories_router)
-app.include_router(auth_router)
-app.include_router(admin_router)
+app.include_router(jokes_router, prefix="/api")
+app.include_router(categories_router, prefix="/api")
+app.include_router(auth_router, prefix="/api")
+app.include_router(admin_router, prefix="/api")
+app.include_router(settings_router, prefix="/api")
+app.include_router(links_router, prefix="/api")
+app.include_router(feedback_router, prefix="/api")
+app.include_router(generate_router, prefix="/api")
+app.include_router(submit_router, prefix="/api/jokes")  # 公开提交接口
 
 
 @app.get("/")

api/main.py.bak

@@ -0,0 +1,39 @@
+from fastapi import FastAPI
+from fastapi.middleware.cors import CORSMiddleware
+
+from app.config import API_TITLE, API_VERSION
+from app.routers import jokes_router, categories_router, auth_router, admin_router, settings_router
+from app.database import Base, engine
+from app.models.setting import AiSetting  # 注册模型，确保 create_all 能看到
+
+# 创建新表（如果不存在）
+Base.metadata.create_all(bind=engine)
+
+# Create FastAPI application
+app = FastAPI(title=API_TITLE, version=API_VERSION)
+
+# Configure CORS
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],
+    allow_credentials=True,
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+# Mount routers
+app.include_router(jokes_router, prefix="/api")
+app.include_router(categories_router, prefix="/api")
+app.include_router(auth_router, prefix="/api")
+app.include_router(admin_router, prefix="/api")
+app.include_router(settings_router, prefix="/api")
+
+
+@app.get("/")
+def root():
+    return {"message": "欢迎使用笑话大全 API"}
+
+
+@app.get("/health")
+def health_check():
+    return {"status": "healthy"}

api/requirements.txt

@@ -1,7 +1,11 @@
 fastapi==0.109.0
 uvicorn[standard]==0.27.0
+gunicorn==21.2.0
 sqlalchemy==2.0.25
 pydantic==2.5.3
 python-jose[cryptography]==3.3.0
 passlib[bcrypt]==1.7.4
-python-multipart==0.0.6
+python-multipart==0.0.6
+openai==1.12.0
+crawl4ai==0.3.0
+httpx==0.27.0

api/server.log

@@ -0,0 +1,501 @@
+INFO:     Will watch for changes in these directories: ['D:\\bwstudio\\joke\\api']
+INFO:     Uvicorn running on http://127.0.0.1:8001 (Press CTRL+C to quit)
+INFO:     Started reloader process [2236] using WatchFiles
+INFO:     Started server process [1824]
+INFO:     Waiting for application startup.
+INFO:     Application startup complete.
+INFO:     127.0.0.1:61949 - "GET /health HTTP/1.1" 200 OK
+INFO:     127.0.0.1:61961 - "POST /api/auth/login HTTP/1.1" 200 OK
+INFO:     127.0.0.1:61964 - "GET /api/admin/jokes?page=1&page_size=3 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62046 - "GET / HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62046 - "GET /favicon.ico HTTP/1.1" 404 Not Found
+INFO:     127.0.0.1:62086 - "GET /api/admin/stats HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62088 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62090 - "GET /api/admin/stats HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62151 - "GET /api/admin/stats HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62154 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62158 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62161 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62164 - "GET /api/admin/jokes/1278 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62167 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62173 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62176 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62179 - "GET /api/admin/jokes/1275 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62182 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62198 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62201 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62204 - "GET /api/admin/jokes/1273 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62208 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62213 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62216 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62219 - "GET /api/admin/jokes/1279 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62223 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62230 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62233 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62236 - "GET /api/admin/jokes/1274 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62239 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62242 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62245 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62261 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62264 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62267 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62270 - "GET /api/admin/jokes/1262 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62273 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62283 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62286 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62289 - "GET /api/admin/jokes/1265 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62292 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62295 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62298 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62302 - "GET /api/admin/settings/active HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62304 - "GET /api/admin/settings HTTP/1.1" 500 Internal Server Error
+ERROR:    Exception in ASGI application
+Traceback (most recent call last):
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\uvicorn\protocols\http\httptools_impl.py", line 419, in run_asgi
+    result = await app(  # type: ignore[func-returns-value]
+             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\uvicorn\middleware\proxy_headers.py", line 84, in __call__
+    return await self.app(scope, receive, send)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\fastapi\applications.py", line 1054, in __call__
+    await super().__call__(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\applications.py", line 123, in __call__
+    await self.middleware_stack(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\middleware\errors.py", line 186, in __call__
+    raise exc
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\middleware\errors.py", line 164, in __call__
+    await self.app(scope, receive, _send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\middleware\cors.py", line 83, in __call__
+    await self.app(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\middleware\exceptions.py", line 62, in __call__
+    await wrap_app_handling_exceptions(self.app, conn)(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\_exception_handler.py", line 64, in wrapped_app
+    raise exc
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\_exception_handler.py", line 53, in wrapped_app
+    await app(scope, receive, sender)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 762, in __call__
+    await self.middleware_stack(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 782, in app
+    await route.handle(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 297, in handle
+    await self.app(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 77, in app
+    await wrap_app_handling_exceptions(app, request)(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\_exception_handler.py", line 64, in wrapped_app
+    raise exc
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\_exception_handler.py", line 53, in wrapped_app
+    await app(scope, receive, sender)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 72, in app
+    response = await func(request)
+               ^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\fastapi\routing.py", line 315, in app
+    content = await serialize_response(
+              ^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\fastapi\routing.py", line 160, in serialize_response
+    return field.serialize(
+           ^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\fastapi\_compat.py", line 147, in serialize
+    return self._type_adapter.dump_python(
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\pydantic\type_adapter.py", line 607, in dump_python
+    return self.serializer.to_python(
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^
+pydantic_core._pydantic_core.PydanticSerializationError: Unable to serialize unknown type: <class 'app.models.setting.AiSetting'>
+INFO:     127.0.0.1:62308 - "GET /api/admin/stats HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62324 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62327 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62330 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62356 - "GET /api/admin/settings/active HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62359 - "GET /api/admin/settings HTTP/1.1" 500 Internal Server Error
+ERROR:    Exception in ASGI application
+Traceback (most recent call last):
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\uvicorn\protocols\http\httptools_impl.py", line 419, in run_asgi
+    result = await app(  # type: ignore[func-returns-value]
+             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\uvicorn\middleware\proxy_headers.py", line 84, in __call__
+    return await self.app(scope, receive, send)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\fastapi\applications.py", line 1054, in __call__
+    await super().__call__(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\applications.py", line 123, in __call__
+    await self.middleware_stack(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\middleware\errors.py", line 186, in __call__
+    raise exc
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\middleware\errors.py", line 164, in __call__
+    await self.app(scope, receive, _send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\middleware\cors.py", line 83, in __call__
+    await self.app(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\middleware\exceptions.py", line 62, in __call__
+    await wrap_app_handling_exceptions(self.app, conn)(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\_exception_handler.py", line 64, in wrapped_app
+    raise exc
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\_exception_handler.py", line 53, in wrapped_app
+    await app(scope, receive, sender)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 762, in __call__
+    await self.middleware_stack(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 782, in app
+    await route.handle(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 297, in handle
+    await self.app(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 77, in app
+    await wrap_app_handling_exceptions(app, request)(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\_exception_handler.py", line 64, in wrapped_app
+    raise exc
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\_exception_handler.py", line 53, in wrapped_app
+    await app(scope, receive, sender)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 72, in app
+    response = await func(request)
+               ^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\fastapi\routing.py", line 315, in app
+    content = await serialize_response(
+              ^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\fastapi\routing.py", line 160, in serialize_response
+    return field.serialize(
+           ^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\fastapi\_compat.py", line 147, in serialize
+    return self._type_adapter.dump_python(
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\pydantic\type_adapter.py", line 607, in dump_python
+    return self.serializer.to_python(
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^
+pydantic_core._pydantic_core.PydanticSerializationError: Unable to serialize unknown type: <class 'app.models.setting.AiSetting'>
+INFO:     127.0.0.1:62362 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62365 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62368 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62371 - "GET /api/admin/stats HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62383 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62386 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62389 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62392 - "GET /api/admin/settings/active HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62395 - "GET /api/admin/settings HTTP/1.1" 500 Internal Server Error
+ERROR:    Exception in ASGI application
+Traceback (most recent call last):
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\uvicorn\protocols\http\httptools_impl.py", line 419, in run_asgi
+    result = await app(  # type: ignore[func-returns-value]
+             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\uvicorn\middleware\proxy_headers.py", line 84, in __call__
+    return await self.app(scope, receive, send)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\fastapi\applications.py", line 1054, in __call__
+    await super().__call__(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\applications.py", line 123, in __call__
+    await self.middleware_stack(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\middleware\errors.py", line 186, in __call__
+    raise exc
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\middleware\errors.py", line 164, in __call__
+    await self.app(scope, receive, _send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\middleware\cors.py", line 83, in __call__
+    await self.app(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\middleware\exceptions.py", line 62, in __call__
+    await wrap_app_handling_exceptions(self.app, conn)(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\_exception_handler.py", line 64, in wrapped_app
+    raise exc
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\_exception_handler.py", line 53, in wrapped_app
+    await app(scope, receive, sender)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 762, in __call__
+    await self.middleware_stack(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 782, in app
+    await route.handle(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 297, in handle
+    await self.app(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 77, in app
+    await wrap_app_handling_exceptions(app, request)(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\_exception_handler.py", line 64, in wrapped_app
+    raise exc
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\_exception_handler.py", line 53, in wrapped_app
+    await app(scope, receive, sender)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 72, in app
+    response = await func(request)
+               ^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\fastapi\routing.py", line 315, in app
+    content = await serialize_response(
+              ^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\fastapi\routing.py", line 160, in serialize_response
+    return field.serialize(
+           ^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\fastapi\_compat.py", line 147, in serialize
+    return self._type_adapter.dump_python(
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\pydantic\type_adapter.py", line 607, in dump_python
+    return self.serializer.to_python(
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^
+pydantic_core._pydantic_core.PydanticSerializationError: Unable to serialize unknown type: <class 'app.models.setting.AiSetting'>
+INFO:     127.0.0.1:62398 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62401 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62404 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62407 - "GET /api/admin/stats HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62414 - "GET /api/admin/stats HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62419 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62422 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62425 - "GET /api/admin/stats HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62429 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62432 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62435 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62438 - "GET /api/admin/stats HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62441 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62444 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62447 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62451 - "GET /api/admin/settings/active HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62453 - "GET /api/admin/settings HTTP/1.1" 500 Internal Server Error
+ERROR:    Exception in ASGI application
+Traceback (most recent call last):
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\uvicorn\protocols\http\httptools_impl.py", line 419, in run_asgi
+    result = await app(  # type: ignore[func-returns-value]
+             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\uvicorn\middleware\proxy_headers.py", line 84, in __call__
+    return await self.app(scope, receive, send)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\fastapi\applications.py", line 1054, in __call__
+    await super().__call__(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\applications.py", line 123, in __call__
+    await self.middleware_stack(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\middleware\errors.py", line 186, in __call__
+    raise exc
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\middleware\errors.py", line 164, in __call__
+    await self.app(scope, receive, _send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\middleware\cors.py", line 83, in __call__
+    await self.app(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\middleware\exceptions.py", line 62, in __call__
+    await wrap_app_handling_exceptions(self.app, conn)(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\_exception_handler.py", line 64, in wrapped_app
+    raise exc
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\_exception_handler.py", line 53, in wrapped_app
+    await app(scope, receive, sender)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 762, in __call__
+    await self.middleware_stack(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 782, in app
+    await route.handle(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 297, in handle
+    await self.app(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 77, in app
+    await wrap_app_handling_exceptions(app, request)(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\_exception_handler.py", line 64, in wrapped_app
+    raise exc
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\_exception_handler.py", line 53, in wrapped_app
+    await app(scope, receive, sender)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 72, in app
+    response = await func(request)
+               ^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\fastapi\routing.py", line 315, in app
+    content = await serialize_response(
+              ^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\fastapi\routing.py", line 160, in serialize_response
+    return field.serialize(
+           ^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\fastapi\_compat.py", line 147, in serialize
+    return self._type_adapter.dump_python(
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\pydantic\type_adapter.py", line 607, in dump_python
+    return self.serializer.to_python(
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^
+pydantic_core._pydantic_core.PydanticSerializationError: Unable to serialize unknown type: <class 'app.models.setting.AiSetting'>
+INFO:     127.0.0.1:62456 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62459 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62462 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62465 - "GET /api/admin/stats HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62468 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62471 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62474 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62477 - "GET /api/admin/settings/active HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62480 - "GET /api/admin/settings HTTP/1.1" 500 Internal Server Error
+ERROR:    Exception in ASGI application
+Traceback (most recent call last):
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\uvicorn\protocols\http\httptools_impl.py", line 419, in run_asgi
+    result = await app(  # type: ignore[func-returns-value]
+             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\uvicorn\middleware\proxy_headers.py", line 84, in __call__
+    return await self.app(scope, receive, send)
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\fastapi\applications.py", line 1054, in __call__
+    await super().__call__(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\applications.py", line 123, in __call__
+    await self.middleware_stack(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\middleware\errors.py", line 186, in __call__
+    raise exc
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\middleware\errors.py", line 164, in __call__
+    await self.app(scope, receive, _send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\middleware\cors.py", line 83, in __call__
+    await self.app(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\middleware\exceptions.py", line 62, in __call__
+    await wrap_app_handling_exceptions(self.app, conn)(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\_exception_handler.py", line 64, in wrapped_app
+    raise exc
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\_exception_handler.py", line 53, in wrapped_app
+    await app(scope, receive, sender)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 762, in __call__
+    await self.middleware_stack(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 782, in app
+    await route.handle(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 297, in handle
+    await self.app(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 77, in app
+    await wrap_app_handling_exceptions(app, request)(scope, receive, send)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\_exception_handler.py", line 64, in wrapped_app
+    raise exc
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\_exception_handler.py", line 53, in wrapped_app
+    await app(scope, receive, sender)
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\starlette\routing.py", line 72, in app
+    response = await func(request)
+               ^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\fastapi\routing.py", line 315, in app
+    content = await serialize_response(
+              ^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\fastapi\routing.py", line 160, in serialize_response
+    return field.serialize(
+           ^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\fastapi\_compat.py", line 147, in serialize
+    return self._type_adapter.dump_python(
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "C:\Users\PF\AppData\Roaming\Python\Python311\site-packages\pydantic\type_adapter.py", line 607, in dump_python
+    return self.serializer.to_python(
+           ^^^^^^^^^^^^^^^^^^^^^^^^^^
+pydantic_core._pydantic_core.PydanticSerializationError: Unable to serialize unknown type: <class 'app.models.setting.AiSetting'>
+INFO:     127.0.0.1:62483 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62486 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62489 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62492 - "GET /api/admin/stats HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62883 - "GET /api/admin/stats HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62885 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62887 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62889 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62892 - "GET /api/admin/jokes/1279 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62895 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62898 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62901 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62904 - "GET /api/admin/jokes/1279 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62907 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62992 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62994 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62997 - "GET /api/jokes/?page=1&page_size=30 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:62999 - "GET /api/jokes/?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63002 - "GET /api/jokes/1275 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63005 - "GET /api/jokes/?page=1&page_size=50 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63008 - "GET /api/jokes/1246 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63011 - "GET /api/jokes/?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63014 - "GET /api/jokes/?page=1&page_size=20&type_ids=1 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63017 - "GET /api/jokes/1279 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63021 - "GET /api/jokes/?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63034 - "GET /api/jokes/?page=1&page_size=20&type_ids=12 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63037 - "GET /api/jokes/?page=1&page_size=20&type_ids=12,1 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63040 - "GET /api/jokes/1279 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63043 - "GET /api/jokes/?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63046 - "GET /api/jokes/1279 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63049 - "GET /api/jokes/?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63052 - "GET /api/jokes/1279 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63055 - "GET /api/jokes/?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63059 - "GET /api/jokes/1277 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63062 - "GET /api/jokes/?page=1&page_size=20 HTTP/1.1" 200 OK
+WARNING:  WatchFiles detected changes in 'app\models\link.py'. Reloading...
+INFO:     127.0.0.1:63122 - "GET /api/jokes/?page=2&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63125 - "GET /api/jokes/?page=3&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63128 - "GET /api/jokes/?page=4&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63131 - "GET /api/jokes/?page=5&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63134 - "GET /api/jokes/?page=7&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63138 - "GET /api/jokes/1140 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63141 - "GET /api/jokes/?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63144 - "GET /api/jokes/?page=1&page_size=20&type_ids=4 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63147 - "GET /api/jokes/923 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63150 - "GET /api/jokes/?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63161 - "GET /api/jokes/?page=1&page_size=20&type_ids=3 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63164 - "GET /api/jokes/?page=1&page_size=20&type_ids=3,4 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63167 - "GET /api/jokes/?page=1&page_size=20&type_ids=3,4,5 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63170 - "GET /api/jokes/?page=1&page_size=20&type_ids=3,4,5,6 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63173 - "GET /api/jokes/?page=1&page_size=20&type_ids=4,5,6 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63176 - "GET /api/jokes/?page=1&page_size=20&type_ids=4,5,6,3 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63179 - "GET /api/jokes/?page=1&page_size=20&type_ids=4,5,6,3,2 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63182 - "GET /api/jokes/?page=1&page_size=20&type_ids=4,5,6,3,2,1 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63185 - "GET /api/jokes/?page=1&page_size=20&type_ids=4,5,6,3,2,1,8 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63188 - "GET /api/jokes/?page=1&page_size=20&type_ids=4,5,6,3,2,1,8,7 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63191 - "GET /api/jokes/?page=1&page_size=20&type_ids=4,5,6,3,2,1,8,7 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63194 - "GET /api/jokes/?page=1&page_size=20&type_ids=4,5,6,3,2,1,8,7 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63197 - "GET /api/jokes/?page=1&page_size=20&type_ids=4,5,6,3,2,1,8,7 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63200 - "GET /api/jokes/?page=1&page_size=20&type_ids=4,5,6,3,2,1,8,7 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63204 - "GET /api/jokes/?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63208 - "GET /api/jokes/1279 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63211 - "GET /api/jokes/?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63214 - "GET /api/jokes/?page=1&page_size=20&type_ids=3 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63217 - "GET /api/jokes/?page=1&page_size=20&type_ids=3,4 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63220 - "GET /api/jokes/?page=1&page_size=20&type_ids=3,4,5 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63223 - "GET /api/jokes/?page=1&page_size=20&type_ids=3,4,5,6 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63226 - "GET /api/jokes/1251 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63229 - "GET /api/jokes/?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63232 - "GET /api/jokes/?page=1&page_size=20&type_ids=2 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63235 - "GET /api/jokes/?page=1&page_size=20&type_ids=2,3 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63238 - "GET /api/jokes/?page=1&page_size=20&type_ids=2,3,4 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63241 - "GET /api/jokes/?page=1&page_size=20&crowd_ids=15 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63244 - "GET /api/jokes/?page=1&page_size=20&crowd_ids=15,16 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63264 - "GET /api/jokes/?page=1&page_size=20&type_ids=2 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63267 - "GET /api/jokes/?page=1&page_size=20&type_ids=2,3 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63270 - "GET /api/jokes/1273 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63273 - "GET /api/jokes/?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63276 - "GET /api/jokes/1277 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63279 - "GET /api/jokes/?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63295 - "GET /api/jokes/?page=2&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63298 - "GET /api/jokes/?page=3&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63301 - "GET /api/jokes/?page=1&page_size=20&type_ids=1 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63304 - "GET /api/jokes/?page=1&page_size=20&type_ids=1,2 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63307 - "GET /api/jokes/?page=1&page_size=20&type_ids=1,2,3 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63310 - "GET /api/jokes/?page=1&page_size=20&type_ids=1,2,3,4 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63313 - "GET /api/jokes/?page=1&page_size=20&type_ids=1,2,3,4,5 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63316 - "GET /api/jokes/?page=1&page_size=20&type_ids=1,2,3,4,5,6 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63319 - "GET /api/jokes/?page=1&page_size=20&type_ids=1,2,3,4,5,6,7 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63322 - "GET /api/jokes/?page=1&page_size=20&type_ids=1,2,3,4,5,6,7,8 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63325 - "GET /api/jokes/?page=1&page_size=20&type_ids=1,2,3,4,5,6,7,8,9 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63328 - "GET /api/jokes/?page=1&page_size=20&type_ids=1,2,3,4,5,6,7,8,9,10 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63331 - "GET /api/jokes/?page=1&page_size=20&type_ids=2,3,4,5,6,7,8,9,10 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63334 - "GET /api/jokes/?page=1&page_size=20&type_ids=2,3,4,5,6,7,8,9 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63337 - "GET /api/jokes/?page=1&page_size=20&type_ids=2,3,4,5,6,7,8 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63340 - "GET /api/jokes/?page=1&page_size=20&type_ids=2,3,4,5,6,7 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63343 - "GET /api/jokes/?page=1&page_size=20&type_ids=2,3,4,5,6 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63346 - "GET /api/jokes/?page=1&page_size=20&type_ids=2,3,4,5 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63349 - "GET /api/jokes/?page=1&page_size=20&type_ids=2,3,4 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63352 - "GET /api/jokes/?page=1&page_size=20&type_ids=2,3 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63369 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63371 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63373 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63376 - "GET /api/admin/jokes?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63379 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63390 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63392 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63394 - "GET /api/jokes/?page=1&page_size=30 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63396 - "GET /api/jokes/?page=1&page_size=20&type_ids=2,3 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63427 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63446 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63448 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63452 - "GET /api/jokes/?page=1&page_size=20&type_ids=2,3 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63450 - "GET /api/jokes/?page=1&page_size=30 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63454 - "GET /api/jokes/1263 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63456 - "GET /api/jokes/?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63466 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63468 - "GET /api/admin/feedbacks HTTP/1.1" 404 Not Found
+INFO:     127.0.0.1:63471 - "GET /api/admin/links HTTP/1.1" 404 Not Found
+INFO:     127.0.0.1:63473 - "GET /api/admin/feedbacks HTTP/1.1" 404 Not Found
+INFO:     127.0.0.1:63476 - "GET /api/admin/links HTTP/1.1" 404 Not Found
+INFO:     127.0.0.1:63479 - "GET /api/admin/feedbacks HTTP/1.1" 404 Not Found
+INFO:     127.0.0.1:63482 - "GET /api/admin/links HTTP/1.1" 404 Not Found
+INFO:     127.0.0.1:63486 - "GET /api/admin/links HTTP/1.1" 404 Not Found
+INFO:     127.0.0.1:63495 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63498 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63503 - "GET /api/jokes/?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63501 - "GET /api/jokes/?page=1&page_size=30 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63508 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63511 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63514 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63517 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63520 - "GET /api/jokes/?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63524 - "GET /api/jokes/?page=1&page_size=30 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63565 - "GET /api/admin/feedbacks HTTP/1.1" 404 Not Found
+INFO:     127.0.0.1:63568 - "GET /api/admin/links HTTP/1.1" 404 Not Found
+INFO:     127.0.0.1:63571 - "GET /api/admin/links HTTP/1.1" 404 Not Found
+INFO:     127.0.0.1:63574 - "GET /api/admin/links HTTP/1.1" 404 Not Found
+INFO:     127.0.0.1:63577 - "GET /api/admin/feedbacks HTTP/1.1" 404 Not Found
+INFO:     127.0.0.1:63580 - "GET /api/admin/links HTTP/1.1" 404 Not Found
+INFO:     127.0.0.1:63583 - "GET /api/admin/feedbacks HTTP/1.1" 404 Not Found
+INFO:     127.0.0.1:63592 - "GET /api/categories/types HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63594 - "GET /api/categories/crowds HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63598 - "GET /api/jokes/?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63596 - "GET /api/jokes/?page=1&page_size=30 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63607 - "GET /api/links HTTP/1.1" 404 Not Found
+INFO:     127.0.0.1:63609 - "GET /api/links HTTP/1.1" 404 Not Found
+INFO:     127.0.0.1:63612 - "GET /api/jokes/?page=1&page_size=20 HTTP/1.1" 200 OK
+INFO:     127.0.0.1:63615 - "GET /api/links HTTP/1.1" 404 Not Found

crawler/__init__.py

@@ -0,0 +1 @@
+"""AI 提示词模板"""

crawler/ai_service.py

@@ -0,0 +1,131 @@
+"""LLM 处理：调用 NVIDIA NIM（OpenAI 兼容 API）进行笑话提取、改写。"""
+
+import json
+import httpx
+from openai import OpenAI
+
+
+class AiService:
+    def __init__(self, api_base: str, username: str, password: str):
+        self.api_base = api_base.rstrip("/")
+        self.username = username
+        self.password = password
+        self.token = None
+        self.client = None
+        self.model = ""
+        self.ai_config = None
+
+    def _login(self) -> str:
+        resp = httpx.post(
+            f"{self.api_base}/api/auth/login",
+            json={"username": self.username, "password": self.password},
+            timeout=30,
+        )
+        resp.raise_for_status()
+        return resp.json()["access_token"]
+
+    def _get(self, path: str) -> dict | list:
+        resp = httpx.get(
+            f"{self.api_base}{path}",
+            headers={"Authorization": f"Bearer {self.token}"},
+            timeout=30,
+        )
+        resp.raise_for_status()
+        return resp.json()
+
+    def setup(self):
+        """登录并从 API 获取 AI 配置（含提示词和温度）"""
+        self.token = self._login()
+        self.ai_config = self._get("/api/admin/settings/active")
+        self.client = OpenAI(
+            base_url=self.ai_config["api_base"],
+            api_key=self.ai_config["api_key"],
+        )
+        self.model = self.ai_config["model_name"]
+
+    def _get_prompt(self, key: str, default: str) -> str:
+        if self.ai_config:
+            val = self.ai_config.get(key)
+            if val and val.strip():
+                return val
+        return default
+
+    def extract_jokes(self, page_content: str, known_types: list[str], known_crowds: list[str]) -> list[dict]:
+        """从页面内容中提取笑话，返回结构化数据。"""
+        system_prompt = self._get_prompt("crawler_extract_prompt",
+            """你是一个笑话提取专家。从给定的网页文本中识别并提取所有笑话、幽默段子或有趣内容。
+要求：
+1. 只返回真正的笑话内容，不要提取普通文章或新闻
+2. 每条笑话需要包含：title（简短标题）、content（完整笑话内容）、type（类型）、crowd（人群）
+3. 如果网页中没有笑话，返回空数组 []
+4. 永远返回合法的 JSON 格式，根节点为数组或包含 jokes 键的对象""")
+
+        user_prompt = f"""网页内容：
+---
+{page_content[:8000]}
+---
+
+已知笑话类型：{', '.join(known_types)}
+已知人群分类：{', '.join(known_crowds)}
+
+请提取所有笑话，以 JSON 格式返回，示例：
+[
+  {{"title": "程序员的幽默", "content": "程序员去相亲...", "types": ["谐音梗", "段子"], "crowds": ["职场", "大学生"]}},
+  {{"title": "...", "content": "...", "types": ["..."], "crowds": ["..."]}}
+]
+
+注意：types 和 crowds 是数组，可以填多个。
+只返回 JSON，不要其他文字。"""
+
+        response = self.client.chat.completions.create(
+            model=self.model,
+            messages=[
+                {"role": "system", "content": system_prompt},
+                {"role": "user", "content": user_prompt},
+            ],
+            temperature=self.ai_config.get("temperature", 0.7) if self.ai_config else 0.7,
+            max_tokens=self.ai_config.get("max_tokens", 2048) if self.ai_config else 2048,
+        )
+
+        raw = response.choices[0].message.content
+        return self._parse_json(raw)
+
+    def rewrite_joke(self, content: str) -> str:
+        """润色单条笑话内容。"""
+        system_prompt = self._get_prompt("crawler_rewrite_prompt",
+            """你是一个幽默作家，负责润色和改写笑话。
+要求：
+1. 保持笑话的核心笑点不变
+2. 语言更通顺、更幽默
+3. 字数控制在原内容的 80%-120% 之间
+4. 不要添加任何解释说明""")
+
+        response = self.client.chat.completions.create(
+            model=self.model,
+            messages=[
+                {"role": "system", "content": system_prompt},
+                {"role": "user", "content": f"请润色以下笑话：\n\n{content}"},
+            ],
+            temperature=0.8,
+            max_tokens=500,
+        )
+        return response.choices[0].message.content.strip()
+
+    def _parse_json(self, raw: str) -> list[dict]:
+        """安全解析 LLM 返回的 JSON。"""
+        try:
+            data = json.loads(raw)
+            if isinstance(data, dict):
+                jokes = data.get("jokes") or data.get("items") or [data]
+            else:
+                jokes = data
+            return [j for j in jokes if isinstance(j, dict) and j.get("content")]
+        except json.JSONDecodeError:
+            if "```json" in raw:
+                raw = raw.split("```json")[1].split("```")[0]
+            elif "```" in raw:
+                raw = raw.split("```")[1].split("```")[0]
+            try:
+                return json.loads(raw.strip())
+            except Exception:
+                return []

crawler/crawler_service.py

@@ -0,0 +1,439 @@
+"""统一网页获取：搜索笑话站点 + 深度翻页抓取。"""
+
+import asyncio
+import os
+import re
+import sys
+from urllib.parse import quote, urljoin, urlparse
+
+# Windows 下禁用 rich 控制台输出，避免 GBK 编码错误
+if sys.platform == "win32":
+    os.environ["PYTHONIOENCODING"] = "utf-8"
+    os.environ["TERM"] = "dumb"
+
+try:
+    from crawl4ai import AsyncWebCrawler
+    from crawl4ai.async_configs import CrawlerRunConfig, BrowserConfig
+    HAS_CRAWL4AI = True
+except ImportError:
+    HAS_CRAWL4AI = False
+    print("[!] crawl4ai 未安装，将使用 requests 替代（不支持 JS 渲染）")
+
+import httpx
+
+
+class CrawlerService:
+    def __init__(self, headless: bool = True):
+        self.headless = headless
+        self.session = None
+        self._crawler = None
+        # 站点级别容错跟踪
+        self.site_failures: dict[str, int] = {}
+        # 最近一次抓取的原始 HTML（供翻页链接发现使用）
+        self._last_raw_html: str | None = None
+
+    def _get_session(self) -> httpx.Client:
+        if self.session is None:
+            self.session = httpx.Client(
+                headers={
+                    "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 Chrome/125.0.0.0 Safari/537.36",
+                    "Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8",
+                    "Accept-Language": "zh-CN,zh;q=0.9,en;q=0.8",
+                    "Referer": "https://www.baidu.com/",
+                },
+                timeout=30,
+                follow_redirects=True,
+            )
+        return self.session
+
+    async def _get_crawler(self):
+        """获取或创建 crawl4ai 实例"""
+        if HAS_CRAWL4AI and self._crawler is None:
+            browser_cfg = BrowserConfig(headless=self.headless, verbose=False)
+            self._crawler = AsyncWebCrawler(config=browser_cfg)
+            await self._crawler.__aenter__()
+            if not self.headless:
+                print("  [*] 浏览器窗口已打开（crawl4ai 控制）")
+        return self._crawler
+
+    async def close(self):
+        if self._crawler:
+            await self._crawler.__aexit__(None, None, None)
+            self._crawler = None
+        if self.session:
+            self.session.close()
+            self.session = None
+
+    # ============================================================
+    # 页面抓取（带重试）
+    # ============================================================
+
+    async def crawl_page_with_retry(self, url: str, max_retries: int = 2, timeout: int = 30000) -> tuple[str, bool]:
+        """抓取页面，返回 (html内容, 是否成功)，失败自动重试"""
+        for attempt in range(max_retries + 1):
+            print(f"  [~] 正在获取页面... (尝试 {attempt+1}/{max_retries+1})")
+            try:
+                html = await self._crawl_one(url, timeout)
+                if html:
+                    print(f"  [~] 页面获取成功，内容长度: {len(html)} 字符")
+                    return html, True
+                print(f"  [!] 页面内容为空")
+            except Exception as e:
+                print(f"  [!] 抓取失败 (尝试 {attempt+1}/{max_retries+1}): {url[:60]} - {e}")
+
+            if attempt < max_retries:
+                wait = 3 * (attempt + 1)
+                print(f"  [~] 等待 {wait} 秒后重试...")
+                await asyncio.sleep(wait)
+        return "", False
+
+    async def _crawl_one(self, url: str, timeout: int) -> str:
+        """单次页面抓取，返回纯文本内容（同时保存原始 HTML 供翻页发现）"""
+        if HAS_CRAWL4AI:
+            # crawl4ai 统一处理 headless / visible 两种模式
+            crawler = await self._get_crawler()
+            result = await crawler.arun(url, config=CrawlerRunConfig(verbose=False))
+            if result and result.success:
+                # fit_html 是 AI 清洗后的正文（翻页链接常被清洗掉）
+                # cleaned_html 保留完整结构（用于翻页发现）
+                self._last_raw_html = result.cleaned_html or result.html or ""
+                html = result.fit_html or result.cleaned_html or ""
+                return self._html_to_text(html)
+            return ""
+        else:
+            # 无 crawl4ai 时 fallback 到 requests
+            resp = self._get_session().get(url, timeout=timeout / 1000)
+            if resp.status_code == 200:
+                self._last_raw_html = resp.text
+                return self._html_to_text(resp.text)
+            return ""
+
+    @staticmethod
+    def _html_to_text(html: str) -> str:
+        """简易 HTML 转纯文本"""
+        # 移除 script/style 标签内容
+        text = re.sub(r'<script[^>]*>.*?</script>', '', html, flags=re.DOTALL | re.IGNORECASE)
+        text = re.sub(r'<style[^>]*>.*?</style>', '', text, flags=re.DOTALL | re.IGNORECASE)
+        # 移除 HTML 标签
+        text = re.sub(r'<[^>]+>', ' ', text)
+        # 合并空白
+        text = re.sub(r'\s+', ' ', text).strip()
+        return text
+
+    # ============================================================
+    # 搜索笑话站点
+    # ============================================================
+
+    # 已知的笑话聚合站域名黑/白名单
+    JOKE_SITE_KEYWORDS = [
+        "笑话大全", "冷笑话", "搞笑段子", "笑话集锦",
+        "幽默笑话", "爆笑笑话", "成人笑话", "小笑话",
+    ]
+
+    async def search_joke_sites(self, keywords: list[str], max_results: int = 10) -> list[dict]:
+        """搜索笑话站点，返回 [{url, title, domain}]"""
+        all_results = []
+
+        for keyword in keywords:
+            results = await self._search_and_filter(keyword, max_results=max_results)
+            all_results.extend(results)
+
+        # 去重（按域名）
+        seen_domains = set()
+        unique = []
+        for r in all_results:
+            domain = r["domain"]
+            if domain not in seen_domains:
+                seen_domains.add(domain)
+                unique.append(r)
+
+        print(f"  [*] 搜索到 {len(unique)} 个唯一站点")
+        for s in unique:
+            print(f"      - {s['domain']}: {s['title'][:40]}")
+        return unique[:max_results]
+
+    async def _search_and_filter(self, keyword: str, max_results: int = 10) -> list[dict]:
+        """搜索并过滤出疑似笑话聚合站的结果"""
+        # 用多个搜索词提高覆盖率
+        search_queries = [
+            f"{keyword} 网站",
+            f"{keyword} 大全",
+            f"{keyword} 列表",
+        ]
+
+        seen = set()
+        sites = []
+
+        for q in search_queries:
+            if len(sites) >= max_results:
+                break
+
+            search_results = await self._search_bing(q, max_pages=2)
+
+            for r in search_results:
+                if len(sites) >= max_results:
+                    break
+
+                url = r["url"]
+                domain = urlparse(url).netloc.lower()
+
+                if domain in seen:
+                    continue
+                seen.add(domain)
+
+                # 过滤：排除已知的单篇文章站点和搜索引擎
+                if self._is_joke_collection_site(url, r.get("title", "")):
+                    r["domain"] = domain
+                    sites.append(r)
+
+        return sites
+
+    def _is_joke_collection_site(self, url: str, title: str) -> bool:
+        """判断URL是否疑似笑话聚合站（不是单篇文章）"""
+        domain = urlparse(url).netloc.lower()
+        path = urlparse(url).path.lower()
+
+        # 排除项
+        exclude_domains = [
+            "bing.com", "microsoft.com", "baidu.com", "google.com",
+            "sohu.com", "sina.com", "163.com", "qq.com", "toutiao.com",
+            "weibo.com", "zhihu.com", "bilibili.com", "douban.com",
+        ]
+        if any(d in domain for d in exclude_domains):
+            return False
+
+        # 排除明显的单篇文章模式
+        single_article_patterns = [
+            r'/p/\d+', r'/article/\d+', r'/post/\d+', r'/archives/\d+',
+            r'/a/\d+', r'/\d{5,}', r'/detail/\d+', r'/read/\d+',
+            r'\.html$',  # 静态 html 文章页
+        ]
+        # 但如果域名本身含 joke 特征，不排除
+        is_joke_domain = any(kw in domain or kw in title for kw in
+                             ["joke", "xiaohua", "笑话", "段子", "幽默", "搞笑"])
+
+        for p in single_article_patterns:
+            if re.search(p, path) and not is_joke_domain:
+                return False
+
+        # 聚合站特征：域名或标题含特定词，或URL有分类/列表模式
+        collection_patterns = [
+            "joke", "xiaohua", "笑话", "段子", "幽默", "搞笑",
+            "/page/", "/list/", "/category/", "/tag/", "joke",
+        ]
+        for p in collection_patterns:
+            if p in domain or p in path or p.lower() in title:
+                return True
+
+        # 有列表/目录模式的也认为是聚合站
+        if re.search(r'(page|list|category|tag|index)', path):
+            return True
+
+        return False
+
+    # ============================================================
+    # 翻页链接发现
+    # ============================================================
+
+    def discover_page_links(self, html: str, base_url: str) -> list[str]:
+        """从页面 HTML 中发现翻页链接，返回去重排序后的 URL 列表"""
+        # 优先使用 _last_raw_html（完整 HTML 而非纯文本），兜底用传入的 html
+        raw = self._last_raw_html or html
+        base_parsed = urlparse(base_url)
+        base_domain = f"{base_parsed.scheme}://{base_parsed.netloc}"
+
+        links = set()
+
+        # 1. <link rel="next">
+        for m in re.finditer(r'<link[^>]*rel="next"[^>]*href="([^"]+)"', raw, re.IGNORECASE):
+            links.add(urljoin(base_domain, m.group(1)))
+
+        # 2. 翻页文字链接（下一页、下页、>、» 等）
+        page_text_patterns = [
+            r'<a[^>]*href="([^"]*page[^"]*)"[^>]*>\s*(?:下一页|下页|下一页»|»|›|>|Next|last)\s*</a>',
+            r'<a[^>]*>\s*(?:下一页|下页|»|›|>)\s*</a>\s*<a[^>]*href="([^"]*)"',
+        ]
+        for pattern in page_text_patterns:
+            for m in re.finditer(pattern, raw, re.IGNORECASE):
+                href = m.group(1).strip()
+                if href and href not in ("#", "javascript:void(0)"):
+                    links.add(urljoin(base_url, href))
+
+        # 3. 提取所有带数字的翻页 link（?page=N, /page/N/, index_N.html, page_N.html, &page=N）
+        page_link_patterns = [
+            r'href="([^"]*[?&]page=(\d+)[^"]*)"',      # ?page=2  &page=2
+            r'href="([^"]*/page/(\d+)[^"]*)"',           # /page/2/
+            r'href="([^"]*[?&]p=(\d+)[^"]*)"',           # ?p=2
+            r'href="([^"]*[?&]pn=(\d+)[^"]*)"',          # ?pn=2
+            r'href="([^"]*[?&]offset=(\d+)[^"]*)"',      # ?offset=10
+            r'href="([^"]*[?&]start=(\d+)[^"]*)"',       # ?start=10
+            r'href="([^"]*[?&]page_index=(\d+)[^"]*)"',  # ?page_index=2
+            # 匹配 xxx_N.html / page_N.html / list_2.html
+            r'href="([^"]*(?:page|list|index)[-_]?(\d+)\.html?)"',
+            # 匹配 /page_N/ 格式
+            r'href="([^"]*/page[-_]?(\d+)/?)"',
+        ]
+        for pattern in page_link_patterns:
+            for m in re.finditer(pattern, raw, re.IGNORECASE):
+                full_url = urljoin(base_url, m.group(1))
+                links.add(full_url)
+
+        # 4. 翻页数字链接
+        for m in re.finditer(r'<a[^>]*href="([^"]*page=(\d+)[^"]*)"[^>]*>\s*\d+\s*</a>', raw, re.IGNORECASE):
+            links.add(urljoin(base_url, m.group(1)))
+
+        # 过滤：只保留同一域名下的链接
+        result = []
+        for link in links:
+            parsed = urlparse(link)
+            if parsed.netloc and parsed.netloc != base_parsed.netloc:
+                continue  # 跨域排除
+            if parsed.path == base_parsed.path and parsed.query == base_parsed.query:
+                continue  # 排除自身
+            result.append(link)
+
+        # 去重排序
+        return sorted(set(result))
+
+    def _extract_page_number(self, url: str) -> int:
+        """从 URL 中提取页码，用于排序"""
+        nums = re.findall(r'page[=/](\d+)|[?&]p=(\d+)|index[-_]?(\d+)|/page[-_]?(\d+)', url, re.IGNORECASE)
+        for n in nums:
+            for g in n:
+                if g:
+                    return int(g)
+        return 99  # 没识别到页码的排最后
+
+    # ============================================================
+    # 分类/标签链接发现
+    # ============================================================
+
+    def discover_category_links(self, html: str, base_url: str) -> list[str]:
+        """从页面 HTML 中发现分类/标签链接，返回去重排序后的 URL 列表"""
+        raw = self._last_raw_html or html
+        base_parsed = urlparse(base_url)
+        base_domain = f"{base_parsed.scheme}://{base_parsed.netloc}"
+
+        links = set()
+
+        # 1. 匹配分类链接（category-N.html, category-N_M.html, tag-N.html 等）
+        cat_patterns = [
+            r'href="([^"]*/(?:category|cat|sort|type)[-_]?\d+(?:[-_]\d+)?\.html?)"',
+            r'href="([^"]*/(?:tag|tags)/?[-_]?\d*)["\s>]',
+            r'href="([^"]*/tags?[-_]?\d+\.html?)"',
+        ]
+        for pattern in cat_patterns:
+            for m in re.finditer(pattern, raw, re.IGNORECASE):
+                full_url = urljoin(base_domain, m.group(1))
+                parsed = urlparse(full_url)
+                # 只保留同域链接
+                if parsed.netloc and parsed.netloc != base_parsed.netloc:
+                    continue
+                links.add(full_url)
+
+        # 2. 过滤：排除单篇文章、首页、搜索页
+        single_article = re.compile(
+            r'/(?:p|post|article|archives|detail|read|xiaohua)/\d+',
+            re.IGNORECASE,
+        )
+        result = []
+        for link in links:
+            parsed = urlparse(link)
+            path = parsed.path.rstrip("/")
+            # 排除自身
+            if path == base_parsed.path.rstrip("/") and parsed.query == base_parsed.query:
+                continue
+            # 排除单篇文章
+            if single_article.search(path):
+                continue
+            # 排除明显的非分类路径（首页翻页）
+            if re.search(r'/page[-_]?\d+\.html?$', path) and 'category' not in path and 'tag' not in path:
+                continue
+            result.append(link)
+
+        return sorted(set(result))
+
+    # ============================================================
+    # Bing 搜索（复用旧逻辑）
+    # ============================================================
+
+    async def _search_bing(self, keyword: str, max_pages: int = 3) -> list[dict]:
+        """搜索 Bing，返回结果 URL 列表"""
+        results = []
+        session = self._get_session()
+
+        for page in range(max_pages):
+            first = page * 10
+            url = f"https://www.bing.com/search?q={quote(keyword)}&first={first}"
+            print(f"  [*] Bing 搜索: {keyword[:20]}")
+
+            html = None
+            try:
+                if HAS_CRAWL4AI:
+                    crawler = await self._get_crawler()
+                    result = await crawler.arun(url, config=CrawlerRunConfig(verbose=False))
+                    if result.success:
+                        html = result.html if result.html else result.cleaned_html
+                else:
+                    resp = session.get(url)
+                    html = resp.text if resp.status_code == 200 else None
+
+                if html:
+                    urls = self._extract_bing_urls(html)
+                    for item in urls:
+                        item["keyword"] = keyword
+                        results.append(item)
+                else:
+                    print(f"  [!] 获取搜索页面失败")
+
+                await asyncio.sleep(2)
+
+            except Exception as e:
+                print(f"  [!] 搜索异常: {e}")
+                continue
+
+        return results
+
+    def _extract_bing_urls(self, html: str) -> list[dict]:
+        """从 Bing 搜索结果 HTML 中提取链接和标题"""
+        results = []
+        pattern = re.compile(
+            r'<h2[^>]*>\s*<a[^>]*href="(https?[^"]+)"[^>]*>(.*?)</a>',
+            re.DOTALL,
+        )
+        for match in pattern.finditer(html):
+            href = match.group(1).strip()
+            title = re.sub(r'<[^>]+>', '', match.group(2)).strip()
+            if href and title and len(title) > 5 and "bing.com" not in href and "microsoft.com" not in href:
+                results.append({"url": href, "title": title})
+        return results
+
+    # 别名兼容
+    async def search_bing(self, keyword: str, max_pages: int = 3) -> list[dict]:
+        return await self._search_bing(keyword, max_pages)
+
+    async def search(self, keyword: str, max_pages: int = 3) -> list[dict]:
+        return await self._search_bing(keyword, max_pages)
+
+    async def search_baidu(self, keyword: str, max_pages: int = 3) -> list[dict]:
+        return await self._search_bing(keyword, max_pages)
+
+    # ============================================================
+    # 旧接口兼容（单页抓取）
+    # ============================================================
+
+    async def crawl_page(self, url: str) -> str:
+        """抓取单个页面（兼容旧接口）"""
+        content, ok = await self.crawl_page_with_retry(url)
+        if ok:
+            return content[:8000]
+        return ""
+
+    async def crawl_batch(self, urls: list[str]) -> list[tuple[str, str]]:
+        """批量抓取（兼容旧接口）"""
+        contents = []
+        for url in urls:
+            content, ok = await self.crawl_page_with_retry(url)
+            contents.append((url, content[:8000] if ok else ""))
+            await asyncio.sleep(1.5)
+        return contents

crawler/main.py

@@ -0,0 +1,85 @@
+"""
+爬虫持续采集入口（深度采集模式）。
+流程：搜索笑话站点 → 翻页深度采集 → AI 提取 → 入库。
+每批采集 20 条后休息 2-10 分钟，出错休息 3-11 分钟后继续。
+按 Ctrl+C 中断。
+
+用法:
+  python crawler/main.py                     # 默认，无头浏览器
+  python crawler/main.py --no-headless       # 显示浏览器窗口（方便测试）
+"""
+import argparse
+import asyncio
+import os
+import sys
+
+if sys.platform == "win32":
+    import shutil
+    os.environ["PYTHONIOENCODING"] = "utf-8"
+    os.environ["TERM"] = "dumb"
+    try:
+        shutil.get_terminal_size()
+    except Exception:
+        pass
+
+sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+
+from crawler.processor import Processor
+
+
+def parse_args():
+    parser = argparse.ArgumentParser(description="笑话爬虫 - 深度采集模式")
+    parser.add_argument(
+        "--no-headless",
+        action="store_true",
+        help="显示浏览器窗口（默认无头模式）",
+    )
+    return parser.parse_args()
+
+
+def main():
+    args = parse_args()
+    headless = not args.no_headless
+
+    api_base = os.getenv("API_BASE", "http://localhost:8001")
+    username = os.getenv("CRAWL_USERNAME", "admin")
+    password = os.getenv("CRAWL_PASSWORD", "admin123")
+    keyword_str = os.getenv("CRAWL_KEYWORD", "笑话大全,搞笑段子,冷笑话,幽默笑话,爆笑笑话,笑话集锦")
+    keywords = [k.strip() for k in keyword_str.split(",") if k.strip()]
+
+    if not keywords:
+        print("错误: 未设置关键词")
+        return
+
+    print(f"=" * 50)
+    print(f"笑话爬虫 - 深度采集模式")
+    print(f"搜索词: {keywords}")
+    print(f"API 地址: {api_base}")
+    print(f"浏览器: {'显示窗口' if not headless else '无头模式'}")
+    print(f"每批 20 条后休息 2-10 分钟")
+    print(f"出错后休息 3-11 分钟后重试")
+    print(f"按 Ctrl+C 终止")
+    print(f"=" * 50)
+
+    try:
+        import crawl4ai
+        print(f"crawl4ai 版本: {crawl4ai.__version__}")
+    except ImportError:
+        print("错误: crawl4ai 未安装，请先运行: pip install crawl4ai")
+        return
+
+    processor = Processor(
+        api_base=api_base,
+        username=username,
+        password=password,
+        headless=headless,
+    )
+
+    try:
+        asyncio.run(processor.run_continuous(keywords=keywords, max_pages=3, batch_size=20))
+    except KeyboardInterrupt:
+        print("\n用户中断，退出")
+
+
+if __name__ == "__main__":
+    main()

crawler/processor.py

@@ -0,0 +1,468 @@
+"""爬虫流程编排：搜索笑话站点 → 深度翻页采集 → AI 提取 → 入库"""
+
+import asyncio
+import hashlib
+import random
+import re
+import httpx
+from urllib.parse import urljoin, urlparse
+
+from crawler.ai_service import AiService
+from crawler.crawler_service import CrawlerService
+
+
+class Processor:
+    def __init__(self, api_base: str, username: str, password: str, headless: bool = True):
+        self.api_base = api_base.rstrip("/")
+        self.username = username
+        self.password = password
+        self.token = None
+        self.ai = None
+        self.crawler = CrawlerService(headless=headless)
+        self.types = []
+        self.crowds = []
+
+    # === API 认证 ===
+    def _login(self) -> str:
+        resp = httpx.post(
+            f"{self.api_base}/api/auth/login",
+            json={"username": self.username, "password": self.password},
+            timeout=30,
+        )
+        resp.raise_for_status()
+        return resp.json()["access_token"]
+
+    def _get(self, path: str) -> dict | list:
+        resp = httpx.get(
+            f"{self.api_base}{path}",
+            headers={"Authorization": f"Bearer {self.token}"},
+            timeout=30,
+        )
+        resp.raise_for_status()
+        return resp.json()
+
+    def _post(self, path: str, data: dict) -> dict:
+        resp = httpx.post(
+            f"{self.api_base}{path}",
+            json=data,
+            headers={"Authorization": f"Bearer {self.token}"},
+            timeout=30,
+        )
+        resp.raise_for_status()
+        return resp.json()
+
+    # === 初始化 ===
+    def setup(self):
+        print("[*] 正在登录...")
+        self.token = self._login()
+        print("[*] 登录成功")
+
+        self.ai = AiService(
+            api_base=self.api_base,
+            username=self.username,
+            password=self.password,
+        )
+        self.ai.setup()
+        print(f"[*] AI 配置: {self.ai.model}")
+
+        self.types = self._get("/api/categories/types")
+        self.crowds = self._get("/api/categories/crowds")
+        print(f"[*] 分类: {len(self.types)} 种类型, {len(self.crowds)} 种人群")
+
+    def _get_existing_hashes(self) -> set[str]:
+        """获取库里已有笑话的 content hash，用于去重"""
+        try:
+            data = self._get("/api/admin/jokes?page=1&page_size=1000")
+            hashes = set()
+            for j in data.get("items", []):
+                content = j.get("content", "")
+                if content:
+                    hashes.add(hashlib.md5(content.encode()).hexdigest())
+            return hashes
+        except Exception:
+            return set()
+
+    def _submit_joke(self, joke: dict) -> bool:
+        """提交单条笑话至 API（status=pending 待审核）"""
+        try:
+            # Support both old format (type/crowd) and new format (types/crowds arrays)
+            type_names = joke.get("types", [joke.get("type", "")])
+            crowd_names = joke.get("crowds", [joke.get("crowd", "")])
+            if isinstance(type_names, str):
+                type_names = [type_names] if type_names else []
+            if isinstance(crowd_names, str):
+                crowd_names = [crowd_names] if crowd_names else []
+
+            type_ids = []
+            crowd_ids = []
+            for n in type_names:
+                for t in self.types:
+                    if t.get("name") == n:
+                        type_ids.append(t.get("id"))
+                        break
+            for n in crowd_names:
+                for c in self.crowds:
+                    if c.get("name") == n:
+                        crowd_ids.append(c.get("id"))
+                        break
+
+            payload = {
+                "title": joke.get("title", "无标题"),
+                "content": joke.get("content", ""),
+                "type_ids": type_ids if type_ids else None,
+                "crowd_ids": crowd_ids if crowd_ids else None,
+                "status": "pending",
+            }
+            self._post("/api/admin/jokes", payload)
+            return True
+        except Exception as e:
+            print(f"  [!] 提交失败: {e}")
+            return False
+
+    # === 主循环 ===
+
+    async def run_continuous(self, keywords: list[str], max_pages: int = 3, batch_size: int = 20):
+        """持续深度采集循环"""
+        print(f"\n>> 深度采集模式启动")
+        print(f"    搜索词: {keywords}")
+        print(f"    每批目标: {batch_size} 条")
+        print(f"    API: {self.api_base}")
+        print(f"    按 Ctrl+C 中断\n")
+
+        self.setup()
+
+        existing_hashes = self._get_existing_hashes()
+        print(f"[*] 当前库中已有 {len(existing_hashes)} 条笑话（用于去重）")
+
+        type_names = [t.get("name", "") for t in self.types]
+        crowd_names = [c.get("name", "") for c in self.crowds]
+        total_saved = 0
+        visited_urls: set[str] = set()
+
+        try:
+            while True:
+                batch_saved = 0
+                print(f"\n{'='*50}")
+                print(f"  开始新一批深度采集 (已累计 {total_saved} 条)")
+                print(f"{'='*50}")
+
+                try:
+                    # Step 1: 搜索笑话站点
+                    print(f"\n[*] 搜索笑话站点...")
+                    sites = await self.crawler.search_joke_sites(keywords, max_results=8)
+
+                    if not sites:
+                        print(f"  [!] 未找到笑话站点，休息后重试")
+                        rest = random.randint(120, 600)
+                        print(f"  休息 {rest//60} 分 {rest%60} 秒...")
+                        await asyncio.sleep(rest)
+                        continue
+
+                    # Step 2: 逐个站点深度采集
+                    for site in sites:
+                        if batch_saved >= batch_size:
+                            break
+
+                        domain = site["domain"]
+                        site_url = site["url"]
+
+                        print(f"\n{'─'*40}")
+                        print(f"  开始采集站点: {domain}")
+                        print(f"{'─'*40}")
+
+                        # 检查该站点失败次数
+                        fail_count = self.crawler.site_failures.get(domain, 0)
+                        if fail_count >= 3:
+                            print(f"  [!] 站点 {domain} 已连续失败 {fail_count} 次，跳过")
+                            continue
+
+                        saved_from_site = await self._crawl_site_deep(
+                            site_url=site_url,
+                            domain=domain,
+                            type_names=type_names,
+                            crowd_names=crowd_names,
+                            existing_hashes=existing_hashes,
+                            visited_urls=visited_urls,
+                            target=batch_size - batch_saved,
+                        )
+                        batch_saved += saved_from_site
+                        total_saved += saved_from_site
+
+                    # Step 3: 休息
+                    rest = random.randint(120, 600)
+                    print(f"\n[OK] 本批入库 {batch_saved} 条，休息 {rest//60} 分 {rest%60} 秒...")
+                    print(f"     按 Ctrl+C 中断\n")
+
+                except Exception as e:
+                    rest = random.randint(180, 660)
+                    print(f"\n[!] 出错: {e}")
+                    print(f"    休息 {rest//60} 分 {rest%60} 秒后重试...")
+
+                await asyncio.sleep(rest)
+
+        except asyncio.CancelledError:
+            print("\n用户中断，退出")
+        finally:
+            await self.crawler.close()
+
+    async def _crawl_site_deep(
+        self,
+        site_url: str,
+        domain: str,
+        type_names: list[str],
+        crowd_names: list[str],
+        existing_hashes: set[str],
+        visited_urls: set[str],
+        target: int,
+    ) -> int:
+        """深度采集一个站点：抓取首页 → 发现翻页链接 → 逐页抓取提取笑话"""
+        saved = 0
+        pages_to_crawl = []
+
+        # 1. 抓取首页
+        print(f"  [*] 抓取首页: {site_url[:60]}")
+        html, ok = await self.crawler.crawl_page_with_retry(site_url)
+        if not ok:
+            self.crawler.site_failures[domain] = self.crawler.site_failures.get(domain, 0) + 1
+            print(f"  [!] 首页抓取失败 ({self.crawler.site_failures[domain]}/3)")
+            return 0
+
+        visited_urls.add(site_url)
+
+        # 2. 从首页提取笑话
+        try:
+            jokes = self.ai.extract_jokes(html[:8000], type_names, crowd_names)
+            saved += self._save_jokes(jokes, existing_hashes, target - saved)
+            print(f"  [+] 首页提取 {len(jokes)} 条，入库 {saved} 条")
+        except Exception as e:
+            print(f"  [!] 首页 AI 提取失败: {e}")
+
+        if saved >= target:
+            self.crawler.site_failures[domain] = 0
+            return saved
+
+        # 3. 发现翻页链接
+        page_links = self.crawler.discover_page_links(html, site_url)
+        # 过滤已访问的链接
+        page_links = [l for l in page_links if l not in visited_urls]
+        # 按页码排序
+        page_links.sort(key=lambda l: self.crawler._extract_page_number(l))
+
+        # 限制翻页深度，避免无限抓取
+        max_pages_per_site = 30
+        page_links = page_links[:max_pages_per_site]
+        print(f"  [*] 发现 {len(page_links)} 个翻页链接，开始逐页采集...")
+
+        # 4. 逐页翻页采集
+        for idx, page_url in enumerate(page_links):
+            if saved >= target:
+                break
+
+            # 检查站点是否已失效
+            if self.crawler.site_failures.get(domain, 0) >= 3:
+                print(f"  [!] 站点 {domain} 失败过多，跳过")
+                break
+
+            print(f"  [*] 翻页 {idx+1}/{len(page_links)}: {page_url[:60]}")
+
+            html, ok = await self.crawler.crawl_page_with_retry(page_url)
+            visited_urls.add(page_url)
+
+            if not ok:
+                self.crawler.site_failures[domain] = self.crawler.site_failures.get(domain, 0) + 1
+                print(f"  [!] 抓取失败 ({self.crawler.site_failures[domain]}/3)")
+                continue
+
+            # 重置失败计数
+            self.crawler.site_failures[domain] = 0
+
+            try:
+                jokes = self.ai.extract_jokes(html[:8000], type_names, crowd_names)
+                new_saved = self._save_jokes(jokes, existing_hashes, target - saved)
+                if new_saved > 0:
+                    saved += new_saved
+                    print(f"  [+] 提取 {len(jokes)} 条，入库 {new_saved} 条 (累计 {saved}/{target})")
+                else:
+                    print(f"  [*] 提取 {len(jokes)} 条（均为重复）")
+            except Exception as e:
+                print(f"  [!] AI 提取失败: {e}")
+
+            await asyncio.sleep(random.uniform(1, 3))
+
+        self.crawler.site_failures[domain] = 0
+
+        # 5. 发现分类链接并逐个深度采集
+        cat_links = self.crawler.discover_category_links(html, site_url)
+        cat_links = [l for l in cat_links if l not in visited_urls]
+        # 限制分类数量
+        max_categories = 20
+        cat_links = cat_links[:max_categories]
+        if cat_links:
+            print(f"  [*] 发现 {len(cat_links)} 个分类链接，开始逐类采集...")
+            for cat_url in cat_links:
+                if saved >= target:
+                    break
+                if self.crawler.site_failures.get(domain, 0) >= 3:
+                    print(f"  [!] 站点 {domain} 失败过多，跳过分类")
+                    break
+
+                saved += await self._crawl_category(
+                    cat_url=cat_url,
+                    domain=domain,
+                    type_names=type_names,
+                    crowd_names=crowd_names,
+                    existing_hashes=existing_hashes,
+                    visited_urls=visited_urls,
+                    target=target - saved,
+                )
+
+        print(f"  [*] 站点 {domain} 采集完成，共入库 {saved} 条")
+        return saved
+
+    async def _crawl_category(
+        self,
+        cat_url: str,
+        domain: str,
+        type_names: list[str],
+        crowd_names: list[str],
+        existing_hashes: set[str],
+        visited_urls: set[str],
+        target: int,
+    ) -> int:
+        """深度采集一个分类页及其翻页"""
+        saved = 0
+        cat_name = cat_url.split("/")[-1].split(".")[0]
+        print(f"\n  {'─'*36}")
+        print(f"  分类采集 [{cat_name}]: {cat_url[:60]}")
+        print(f"  {'─'*36}")
+
+        # 1. 抓取分类首页
+        html, ok = await self.crawler.crawl_page_with_retry(cat_url)
+        visited_urls.add(cat_url)
+        if not ok:
+            self.crawler.site_failures[domain] = self.crawler.site_failures.get(domain, 0) + 1
+            print(f"  [!] 分类首页抓取失败 ({self.crawler.site_failures[domain]}/3)")
+            return 0
+
+        # 2. AI 提取笑话
+        try:
+            jokes = self.ai.extract_jokes(html[:8000], type_names, crowd_names)
+            saved += self._save_jokes(jokes, existing_hashes, target - saved)
+            print(f"  [+] 分类首页提取 {len(jokes)} 条，入库 {saved} 条")
+        except Exception as e:
+            print(f"  [!] 分类首页 AI 提取失败: {e}")
+
+        if saved >= target:
+            return saved
+
+        # 3. 发现该分类的翻页链接
+        # 先尝试通用翻页模式，再尝试分类特定翻页（category-5_2.html）
+        page_links = self.crawler.discover_page_links(html, cat_url)
+        # 从当前分类 URL 派生出分类翻页模式（e.g. category-5 → category-5_2.html）
+        cat_base = cat_url.split("/")[-1].replace(".html", "")
+        cat_page_pattern = re.compile(
+            rf'href="([^"]*{re.escape(cat_base)}[-_]?(\d+)\.html?)"',
+            re.IGNORECASE,
+        )
+        raw = self.crawler._last_raw_html or html
+        for m in cat_page_pattern.finditer(raw):
+            full_url = urljoin(cat_url, m.group(1))
+            page_links.append(full_url)
+
+        page_links = [l for l in page_links if l not in visited_urls]
+        page_links = list(set(page_links))  # 去重
+        page_links.sort(key=lambda l: self.crawler._extract_page_number(l))
+
+        # 限制翻页深度
+        max_pages_per_cat = 20
+        page_links = page_links[:max_pages_per_cat]
+        if page_links:
+            print(f"  [*] 发现 {len(page_links)} 个翻页链接，开始逐页采集...")
+
+        # 4. 逐页抓取
+        for idx, page_url in enumerate(page_links):
+            if saved >= target:
+                break
+            if self.crawler.site_failures.get(domain, 0) >= 3:
+                print(f"  [!] 站点 {domain} 失败过多，跳过本分类")
+                break
+
+            print(f"  [*] 翻页 {idx+1}/{len(page_links)}: {page_url[:60]}")
+
+            html, ok = await self.crawler.crawl_page_with_retry(page_url)
+            visited_urls.add(page_url)
+
+            if not ok:
+                self.crawler.site_failures[domain] = self.crawler.site_failures.get(domain, 0) + 1
+                print(f"  [!] 分类翻页抓取失败 ({self.crawler.site_failures[domain]}/3)")
+                continue
+
+            self.crawler.site_failures[domain] = 0
+
+            try:
+                jokes = self.ai.extract_jokes(html[:8000], type_names, crowd_names)
+                new_saved = self._save_jokes(jokes, existing_hashes, target - saved)
+                if new_saved > 0:
+                    saved += new_saved
+                    print(f"  [+] 提取 {len(jokes)} 条，入库 {new_saved} 条 (累计 {saved}/{target})")
+                else:
+                    print(f"  [*] 提取 {len(jokes)} 条（均为重复）")
+            except Exception as e:
+                print(f"  [!] AI 提取失败: {e}")
+
+            await asyncio.sleep(random.uniform(1, 3))
+
+        print(f"  [*] 分类 [{cat_name}] 采集完成，入库 {saved} 条")
+        return saved
+
+    def _save_jokes(self, jokes: list[dict], existing_hashes: set[str], limit: int) -> int:
+        """去重并入库笑话，返回成功入库数"""
+        saved = 0
+        for joke in jokes:
+            if saved >= limit:
+                break
+            content_text = joke.get("content", "")
+            if not content_text:
+                continue
+            h = hashlib.md5(content_text.encode()).hexdigest()
+            if h in existing_hashes:
+                continue
+            existing_hashes.add(h)
+            if self._submit_joke(joke):
+                saved += 1
+                print(f"  [+] 入库: {joke.get('title', '')[:30]}")
+        return saved
+
+    # === 单站点采集（供 site_crawler.py 调用） ===
+
+    async def crawl_site(self, site_url: str, batch_size: int = 9999):
+        """初始化后深度采集单个站点"""
+        print(f"\n>> 单站点采集: {site_url}\n")
+        self.setup()
+
+        existing_hashes = self._get_existing_hashes()
+        print(f"[*] 当前库中已有 {len(existing_hashes)} 条笑话（用于去重）")
+
+        type_names = [t.get("name", "") for t in self.types]
+        crowd_names = [c.get("name", "") for c in self.crowds]
+        domain = urlparse(site_url).netloc.lower()
+        visited_urls: set[str] = set()
+
+        try:
+            saved = await self._crawl_site_deep(
+                site_url=site_url,
+                domain=domain,
+                type_names=type_names,
+                crowd_names=crowd_names,
+                existing_hashes=existing_hashes,
+                visited_urls=visited_urls,
+                target=batch_size,
+            )
+            print(f"\n[OK] 站点采集完成，共入库 {saved} 条笑话")
+        finally:
+            await self.crawler.close()
+
+    # === 旧接口兼容 ===
+    async def run(self, keywords: list[str], max_pages: int = 3):
+        """单轮爬取（旧接口，内部调用 run_continuous）"""
+        await self.run_continuous(keywords, max_pages, batch_size=9999)

crawler/site_crawler.py

@@ -0,0 +1,165 @@
+"""
+单站点深度采集工具。
+从 JSON 文件按编号加载站点，或直接指定 URL，深度采集该站点所有笑话。
+
+用法:
+  python crawler/site_crawler.py --id 1                    # 采集 site_finder 发现的 #1 站点
+  python crawler/site_crawler.py --id 1,2,3               # 批量采集多个站点
+  python crawler/site_crawler.py --url https://...         # 直接采集指定 URL
+  python crawler/site_crawler.py --id 1 --no-headless      # 显示浏览器窗口
+"""
+import argparse
+import asyncio
+import json
+import os
+import sys
+
+if sys.platform == "win32":
+    import shutil
+    os.environ["PYTHONIOENCODING"] = "utf-8"
+    os.environ["TERM"] = "dumb"
+    try:
+        shutil.get_terminal_size()
+    except Exception:
+        pass
+
+sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+
+from crawler.processor import Processor
+
+
+def parse_args():
+    parser = argparse.ArgumentParser(description="单站点深度采集工具")
+    parser.add_argument(
+        "--id",
+        type=str,
+        default=None,
+        help="站点编号（从 site_finder 生成的 JSON 读取），多个用逗号分隔",
+    )
+    parser.add_argument(
+        "--url",
+        type=str,
+        default=None,
+        help="直接指定站点 URL（与 --id 二选一）",
+    )
+    parser.add_argument(
+        "--sites-file",
+        type=str,
+        default="joke_sites.json",
+        help="站点列表 JSON 文件路径（默认: joke_sites.json）",
+    )
+    parser.add_argument(
+        "--api-base",
+        type=str,
+        default=os.getenv("API_BASE", "http://localhost:8001"),
+        help="API 服务地址",
+    )
+    parser.add_argument(
+        "--username",
+        type=str,
+        default=os.getenv("CRAWL_USERNAME", "admin"),
+        help="管理员用户名",
+    )
+    parser.add_argument(
+        "--password",
+        type=str,
+        default=os.getenv("CRAWL_PASSWORD", "admin123"),
+        help="管理员密码",
+    )
+    parser.add_argument(
+        "--no-headless",
+        action="store_true",
+        help="显示浏览器窗口",
+    )
+    return parser.parse_args()
+
+
+def load_sites(path: str) -> list[dict]:
+    """从 JSON 文件加载站点列表"""
+    if not os.path.exists(path):
+        print(f"错误: 站点文件 {path} 不存在，请先运行 site_finder.py")
+        sys.exit(1)
+    try:
+        with open(path, "r", encoding="utf-8") as f:
+            data = json.load(f)
+        if not data:
+            print(f"错误: 站点文件 {path} 为空")
+            sys.exit(1)
+        return data
+    except Exception as e:
+        print(f"错误: 读取站点文件失败: {e}")
+        sys.exit(1)
+
+
+def resolve_sites(args) -> list[str]:
+    """解析 --id 或 --url 参数，返回待采集的 URL 列表"""
+    if args.url:
+        return [args.url]
+
+    if not args.id:
+        print("错误: 请指定 --id 或 --url")
+        print("  例如: python crawler/site_crawler.py --id 1")
+        print("  例如: python crawler/site_crawler.py --url https://xiaohua.com")
+        sys.exit(1)
+
+    # 解析编号列表 "1,2,3" → [1, 2, 3]
+    try:
+        ids = [int(x.strip()) for x in args.id.split(",") if x.strip()]
+    except ValueError:
+        print("错误: --id 参数必须是数字，多个用逗号分隔")
+        sys.exit(1)
+
+    sites = load_sites(args.sites_file)
+    found = []
+    for sid in ids:
+        match = [s for s in sites if s["id"] == sid]
+        if match:
+            found.append(match[0])
+            print(f"  [*] 站点 #{sid}: {match[0]['domain']} — {match[0]['title'][:40]}")
+        else:
+            print(f"  [!] 站点 #{sid} 未找到（可用编号: {[s['id'] for s in sites[:10]]}...)")
+
+    if not found:
+        print("错误: 没有找到有效的站点编号")
+        sys.exit(1)
+
+    return [s["url"] for s in found]
+
+
+def main():
+    args = parse_args()
+    headless = not args.no_headless
+    urls = resolve_sites(args)
+
+    print(f"=" * 50)
+    print(f"单站点深度采集工具")
+    print(f"目标站点: {len(urls)} 个")
+    for u in urls:
+        print(f"  - {u}")
+    print(f"API 地址: {args.api_base}")
+    print(f"浏览器: {'显示窗口' if not headless else '无头模式'}")
+    print(f"=" * 50)
+
+    try:
+        import crawl4ai
+        print(f"crawl4ai 版本: {crawl4ai.__version__}")
+    except ImportError:
+        print("错误: crawl4ai 未安装，请先运行: pip install crawl4ai")
+        return
+
+    processor = Processor(
+        api_base=args.api_base,
+        username=args.username,
+        password=args.password,
+        headless=headless,
+    )
+
+    try:
+        for url in urls:
+            asyncio.run(processor.crawl_site(url))
+    except KeyboardInterrupt:
+        print("\n用户中断，退出")
+
+
+if __name__ == "__main__":
+    main()