chore: gitignore 忽略前端测试题图截图

上轮 3 次 commit 只入库了修补文件,主体目录漏入库。本次补入 backend/app/、backend/requirements.txt、backend/scripts/__init__.py。同步在 .gitignore 加 homePC/(制片人家用机带过来的参考文件,不进库)。

.clinerules

@@ -2,8 +2,8 @@
 
 > 项目:TPS 中台(央视《军事科技》栏目内部工作台)
 > 仓库:`tps-dashboard`(Gitea,团队可见)
-> 修订版本:v3 · 2026-05-06
-> 适用对象:Cline + Sonnet(Plan)+ DeepSeek pro(Act),所有协作者必读
+> 修订版本:v5 · 2026-05-15
+> 适用对象:Cline + MiniMax M2.7(Plan + Act),所有协作者必读
 
 ---
 
@@ -25,33 +25,23 @@
 
 | 角色 | 模型 / 工具 | 职责 |
 |---|---|---|
-| 制片人 | 刘统(人) | 拍板、审核、领域知识权威、视觉把关 |
+| 制片人 | 刘通(人) | 拍板、审核、领域权威、视觉把关 |
 | 高级技术顾问 | Claude Opus 4.7(claude.ai 网页) | 架构指导、方案审核、风险预警、Cline 卡壳兜底 |
-| **架构师(Phase 0-1)** | **Sonnet 4.6**(Cline Plan 模式) | 项目内方案设计、Act 改动审查 |
-| **架构师(Phase 2 起)** | **MiniMax M2.7**(Cline Plan 模式)| **Sonnet 余额耗尽后接班 Plan,API Key 已 ready** |
-| 执行者 | DeepSeek v4 pro(Cline Act 模式) | 实际写代码,**全程不变** |
+| 架构师 | MiniMax M2.7(Cline Plan 模式) | 项目内方案设计、Act 改动审查 |
+| 执行者 | MiniMax M2.7(Cline Act 模式) | 实际写代码 |
 | 备选执行者 | DeepSeek v4 flash | 仅日常小改(改字段、加字段、改文案) |
 
-**模型阶段性安排(2026-05-06 拍板)**:
+**当前模型配置(2026-05-14 拍板)**:
 
-- **Phase 0-1(骨架期)**:**Plan = Sonnet 4.6**,Act = DeepSeek pro
-- **Phase 2 起**:**Plan = MiniMax M2.7**(标准版,非 highspeed),Act 仍 DeepSeek pro
-- **切换的核心动机**:Sonnet 单次调用昂贵且 Plan 模式调用频次最高,换 MiniMax M2.7 后单次成本降至 1/50
-- DeepSeek 全程在岗,不退役
+- **Plan 模式**:MiniMax M2.7(标准版,非 highspeed)
+- **Act 模式**:MiniMax M2.7
+- 家用机已退役,唯一开发环境为单位 4090D(E:\tps-dashboard)
+- API 配置:`base_url = https://api.minimaxi.com/v1`,Model ID = `MiniMax-M2.7`
 
-**MiniMax M2.7 接入信息(配置在 Cline 的 Plan 模式)**:
+**模型降级口诀**:
 
-- 官方 API:`base_url = https://api.minimaxi.com/v1`(注意是 minimaxi)
-- Model ID:`MiniMax-M2.7`(标准版)
-- **不用** highspeed 版(贵一倍,Plan 模式不需要那点速度)
-- API Key 已由制片人申请就绪
-
-**模型切换的判断口诀**(Phase 2 起):
-
-- Plan 出方案、审 Act 改动 → **MiniMax M2.7**
-- Act 写新代码 / 建结构 / 改架构 → **DeepSeek pro**
-- Act 改小 bug / 加字段 / 改文案 → **DeepSeek flash**(可选)
-- Plan 连续两次方案质量明显下降 → **临时切回 Sonnet 顶住**(允许)
+- Plan 连续两次方案质量明显下降 → **临时切回 claude.ai 找 Opus 4.7 重新规划,不在 Cline 里换模型**
+- Act 连续两次实现质量明显下降 → **优先切回 claude.ai 找 Opus 4.7 审 Plan,看是不是任务拆得太大;如果是小改任务,可临时切到备选执行者 DeepSeek v4 flash;不在 Cline 里硬撑**
 
 ---
 
@@ -59,8 +49,8 @@
 
 ### 3.1 Plan → Act → 完工三阶段
 
-1. **Plan 阶段**:Sonnet 出方案,**复杂方案先发 Opus 4.7 审一遍**,审过后再切 Act
-2. **Act 阶段**:DeepSeek pro 严格按 Plan 实施,**禁止自作主张扩展范围**
+1. **Plan 阶段**:MiniMax M2.7 出方案,**复杂方案先发 Opus 4.7 审一遍**,审过后再切 Act
+2. **Act 阶段**:MiniMax M2.7 严格按 Plan 实施,**禁止自作主张扩展范围**
 3. **完工阶段**:Act 必须给出"完工报告",列出所有改动文件、设计决策、潜在风险
 
 ### 3.2 哪些事 Act 必须停下让 Plan 审
@@ -120,7 +110,7 @@ Phase 内的子任务、修复独立 bug、前端骨架、后端骨架——都
 
 ### 5.2 年度目标只增不改(时间不变性)
 
-`yearly_targets` 表每年新增一行,**历史年份的目标行绝不可被覆盖或删除**。
+`yearly_targets` 表每年新增一行,**历史年份的目标行绝不可被覆盖或删除**:
 
 - 前端展示某期数据时,**根据该期播出年份的目标判颜色**,不是"当前年份的目标"
 - 2024 年播的节目永远按 2024 年的 0.6448 / 0.8989 判,2026 年回看也不会变色
@@ -134,7 +124,7 @@ SQL 端:`yearly_targets.year` 加 UNIQUE 约束,代码端 INSERT 时用 `ON CONF
 
 - **在职编导**:两字段都填(冗余但稳)
 - **离职编导**:`editor_id` 留空(NULL),只填 `editor_name_snapshot`(如"张颖")
-- **历史导入**:按姓名匹配在职用户,匹配不上只填快照
+- **历史导入**:按姓名匹配在职用户,匹配不上只填快照,不报错不丢数据
 - **查询某编导历史作品**:**先按 editor_id,失败回退按 editor_name_snapshot 文本匹配**
 
 DeepSeek 容易写"editor_id 外键非空 + 删人级联"——**禁止**。一旦张颖离职,她做过的几十期节目就全成"无主"了,这是栏目最不能接受的。
@@ -157,6 +147,11 @@ AI Prompt 模板必须默认走双段、备选纯叙述。docxtpl 模板设计
 | `zebian` | 责编 | 维护仪表盘 / 知识库 / 排期 |
 | `biandao` | 编导 | 使用 TPS / 查看知识库 / 生成报题单 / 个人热点雷达 |
 
+**yearly_targets 写权限收紧(2026-05-20)**:
+- POST/PATCH: zhipianren + zebian 可写,biandao 返回 403
+- DELETE: 仅 zhipianren 可操作,zebian 和 biandao 返回 403
+- GET: 三角色均可读
+
 **所有 SQL CHECK、SQLModel Enum、API 鉴权、前端权限判断、菜单显示**——一律使用拼音值,**不得替换为英文**。
 
 ---
@@ -350,7 +345,7 @@ allow_origins=["http://localhost:5173"]   # 必须显式列,不能用 ["*"]
 
 **单位机器 Phase 1 后端可选路径**:
 
-- **路径 A**:从 0 重写(Sonnet + DeepSeek 重新生成)
+- **路径 A**:从 0 重写(MiniMax M2.7 重新生成所有后端文件)
 - **路径 B**(推荐):用户把家用机的 backend 目录打包 zip 拷到单位机,解压进新仓库,Cline 跑一遍跑通即可
 
 由制片人在 Phase 1 启动时拍板选 A 或 B。
@@ -376,11 +371,26 @@ Cline 内的 Sonnet/DeepSeek 不直接与 Opus 通信,流程如下:
 |---|---|---|
 | v1 | 2026-04 | 初版,定模型分工与栏目业务约定 |
 | v2 | 2026-05-04 | 加技术栈精简版、Phase 拆分 |
-| **v3** | **2026-05-06** | **加 Git 工作流、已知避坑清单、单位机重启路径;MiniMax M2.7 Phase 2 起接班 Plan 模式(替代 Sonnet),DeepSeek pro 全程在岗** |
+| v3 | 2026-05-06 | 加 Git 工作流、已知避坑清单、单位机重启路径;MiniMax M2.7 Phase 2 起接班 Plan 模式(替代 Sonnet),DeepSeek pro 全程在岗 |
+| v4 | 2026-05-14 | 家用机退役,唯一开发环境为单位 4090D(E:\tps-dashboard);Phase 2 起 MiniMax M2.7 正式日常使用;config.py 改为 dotenv 直接加载,pydantic-settings 从 requirements.txt 移除 |
+| **v5** | **2026-05-15** | **合入第十三章"Opus 4.7 起草 task brief 的节制原则";Phase 2 正式开工** |
 
 **修订原则**:.clinerules 是宪法不是变更日志,改动应慎重。临时执行细节进 phase{N}_log.md,不进本文件。
 
 ---
 
+## 十三、Opus 4.7 起草 task brief 的节制原则
+
+Opus 4.7 给 MiniMax 写 task brief 时遵守:
+
+- **一屏内为目标**(约 60-80 行,不超过 100 行)
+- **写四类东西**:任务范围、核心约束、业务/纪律红线、验收清单
+- **不写四类东西**:代码片段、目录结构细节、变量名、字段命名建议
+- 把"怎么实现"留给 MiniMax Plan 模式自己长出来,Opus 只审 Plan 不替 Plan
+- 自检口诀:这条是品味判断还是代码细节?品味判断留,代码细节砍
+- 例外:业务红线里涉及具体字段名(如 `audience_share` / `editor_id` / `editor_name_snapshot` / `yearly_targets.year`)必须点名,因为写错字段名就是写错业务
+
+---
+
 *本文件最高优先级。任何冲突以本文件为准。*
 *下次修订建议时机:Phase 2 末(评估 MiniMax 切换、整理 Phase 2 经验)。*

.gitignore

@@ -13,23 +13,39 @@ dist/
 build/
 .vite/
 
-# 环境配置(永远不上传)
+# 莽聨呕暮藰聝茅聟聧莽藵沤(膰掳赂膷偶聹盲赂聧盲赂聤盲藕聽)
 .env
 .env.local
 
-# 编辑器
+# 莽藕聳膷啪聭暮聶篓
 .vscode/
 .idea/
 *.swp
 
-# 操作系统
+# 膰聯聧盲藵聹莽艂钮莽钮聼
 .DS_Store
 Thumbs.db
 
-# 日志运行产物
+# 膰聴慕暮偶聴膷偶聬膷膭聦盲艧搂莽聣艩
 *.log
 cookie.txt
+cookies*.txt
 
-# 临时文件
+# 盲赂麓膰聴艣膰聳聡盲钮艣
 *.tmp
 *.bak
+homePC/
+
+# Syncthing (膰聻聛莽艩艧茅聴麓 NAS 暮聬聦膰颅慕暮藝慕暮聟藝盲艧搂莽聣艩)
+.stignore
+
+# 臉媒啪脻偶芒膮赂藝脻 / SQL dump(藳钮藵艡掳膰膮啪偶芒)
+backups/
+
+# 鏁忔劅鑴氭湰(姘镐笉涓婁紶)
+backend/scripts/reset_password.py
+  
+# 前端测试/占位图片(题图上传测试产物)  
+frontend/src/assets/*.jpg  
+frontend/src/assets/*.jpeg  
+frontend/src/assets/*.png 

README.md

@@ -10,7 +10,7 @@
 一个面向《军事科技》栏目 8 人团队的内部工作台,核心模块包括:
 
 - **TPS**(选题策划系统):查重 + 知识库参考 + 报题单一键生成
-- **仪表盘**:近 4 期收视成绩 + 未来 6 期排期一目了然
+- **仪表盘**:近 9 期收视成绩 + 未来 6 期排期一目了然
 - **知识库**:往期文稿、军报材料、专家库、外拍线索一站式检索
 - **个人首页**:每位编导有自己的工作台 + 今日热点雷达
 - **拍摄资源库**:外拍联系人 / 单位 / 线索来源结构化沉淀
@@ -47,10 +47,11 @@
 
 | Phase | 状态 |
 |---|---|
-| Phase 0 环境 + Git | 单位机进行中 |
-| Phase 1 后端骨架 | 家用机已完成,单位机走"路径 B 迁移" |
-| Phase 1 前端骨架 | 未开始 |
-| Phase 2-5 | 未开始 |
+| Phase 0 环境 + Git | ✅ 已完成 |
+| Phase 1 后端骨架 + 前端骨架 | ✅ 已完成 |
+| Phase 2 仪表盘 + 批量导入 + 写权限收紧 | ✅ 已完成 |
+| Phase 3 知识库基础设施 | 进行中 — 已完成至 Task 3（知识库树形视图按来源分组），语义搜索尚未做 |
+| Phase 4–5 | 未开始 |
 
 ---
 
@@ -58,15 +59,35 @@
 
 | 角色 | 谁 / 什么 | 做什么 |
 |---|---|---|
-| 制片人 | 刘统(人) | 拍板、产品决策、领域知识 |
+| 制片人 | 刘通(人) | 拍板、产品决策、领域知识 |
 | 高级技术顾问 | Claude Opus 4.7(claude.ai 网页) | 架构指导、关键决策审核 |
-| 架构师 | Sonnet 4.6 → MiniMax M2.7(Phase 2 起) | Cline Plan 模式,出方案 |
-| 执行者 | DeepSeek v4 pro | Cline Act 模式,写代码 |
+| 架构师 | MiniMax M2.7 | Cline Plan 模式,出方案 |
+| 执行者 | MiniMax M2.7 | Cline Act 模式,写代码 |
+| 备选执行者 | DeepSeek v4 flash | 仅日常小改(改字段、加字段、改文案) |
 
 详见 `.clinerules` 第二章。
 
 ---
 
+## 角色权限表
+
+> 数据来源:`.clinerules` 5.5 节 + `project_plan.md` 第四节
+
+| 资源 / 操作 | zhipianren(制片人) | zebian(责编) | biandao(编导) |
+|---|---|---|---|
+| **yearly_targets** | | | |
+| GET 列表/详情 | ✅ | ✅ | ✅ |
+| POST 新增 | ✅ | ✅ | ❌ 403 |
+| PATCH 修改 | ✅ | ✅ | ❌ 403 |
+| DELETE 删除 | ✅ | ❌ 403 | ❌ 403 |
+| **episodes** | | | |
+| GET | ✅ | ✅ | ✅ |
+| POST/PATCH/DELETE | ✅ | ✅ | 受限(见 API 文档) |
+
+> 规则:.clinerules 5.5 角色枚举锁定拼音值(zhipianren / zebian / biandao),不许新增枚举。
+
+---
+
 ## 技术栈速览
 
 - 前端:React 18 + Vite + Ant Design 5 + Zustand(JavaScript,不用 TS)
@@ -153,10 +174,10 @@ uvicorn app.main:app --reload
 
 | 状况 | 找谁 |
 |---|---|
-| 项目方向 / 产品决策 | 刘统(制片人) |
+| 项目方向 / 产品决策 | 刘通(制片人) |
 | 关键架构 / 风险 / 卡壳兜底 | Claude Opus 4.7(claude.ai 网页) |
-| 日常方案 / 改动审查 | Sonnet 4.6(VS Code Cline Plan) |
-| 写代码 | DeepSeek v4 pro(VS Code Cline Act) |
+| 日常方案 / 改动审查 | MiniMax M2.7(VS Code Cline Plan) |
+| 写代码 | MiniMax M2.7(VS Code Cline Act) |
 | Git 历史搞乱了 / 不小心 commit 了 .env | **立刻找 Claude Opus 4.7,不要自己瞎搞** |
 
 ---
@@ -166,6 +187,8 @@ uvicorn app.main:app --reload
 | 版本 | 日期 | 变化 |
 |---|---|---|
 | v1 | 2026-05-06 | 初版,作为单位机仓库门牌 |
+| **v2** | **2026-05-14** | **家用机退役,唯一开发环境为单位 4090D;Phase 2 起 MiniMax M2.7 正式日常使用;更新版本演进记录** |
+| **v3** | **2026-05-27** | **进度校准:Phase 1/2 标为完成;Phase 3 Task 3 知识库树形视图已完成;清理"家用机/路径B"历史表述,不再保留迁移叙事** |
 
 ---
 

backend/app/api/auth.py

@@ -0,0 +1,46 @@
+"""
+认证路由 — 登录 / 登出 / 当前用户
+"""
+
+from fastapi import APIRouter, Depends, HTTPException, Request, status
+from sqlmodel import Session, select
+
+from app.core.deps import get_current_user
+from app.core.security import verify_password
+from app.db.session import get_session
+from app.models.user import User
+from app.schemas.auth import LoginRequest, UserResponse
+
+router = APIRouter(prefix="/api/auth", tags=["认证"])
+
+
+@router.post("/login", response_model=UserResponse)
+def login(body: LoginRequest, request: Request, session: Session = Depends(get_session)):
+    """账号登录。验证成功后写入 session，返回用户信息。"""
+    statement = select(User).where(User.username == body.username)
+    user = session.exec(statement).first()
+    if user is None or not user.is_active:
+        raise HTTPException(
+            status_code=status.HTTP_401_UNAUTHORIZED,
+            detail="用户名或密码错误",
+        )
+    if not verify_password(body.password, user.password_hash):
+        raise HTTPException(
+            status_code=status.HTTP_401_UNAUTHORIZED,
+            detail="用户名或密码错误",
+        )
+    request.session["user_id"] = user.id
+    return user
+
+
+@router.post("/logout")
+def logout(request: Request):
+    """退出登录，清空当前 session。"""
+    request.session.clear()
+    return {"message": "已退出登录"}
+
+
+@router.get("/me", response_model=UserResponse)
+def get_me(user: User = Depends(get_current_user)):
+    """获取当前登录用户的信息。"""
+    return user

backend/app/api/dashboard.py

@@ -0,0 +1,121 @@
+"""
+仪表盘 API — 题图上传 / 查询
+
+权限：
+- POST /cover — 仅 zhipianren / zebian（biandao 返回 403）
+- GET /cover  — 三角色均可读
+"""
+
+import os
+import uuid
+from datetime import datetime
+from pathlib import Path
+
+from fastapi import APIRouter, Depends, HTTPException, UploadFile, File, Form, status
+from sqlmodel import Session, select
+
+from app.core.deps import get_current_user, require_role
+from app.db.session import get_session
+from app.models.user import User, UserRole
+from app.models.cover_settings import CoverSettings
+
+router = APIRouter(prefix="/api/dashboard", tags=["仪表盘"])
+
+# 题图文件存放目录（相对于 backend/）
+STATIC_COVERS_DIR = Path(__file__).parent.parent.parent / "static" / "covers"
+STATIC_COVERS_URL = "/static/covers"
+
+# 确保目录存在
+STATIC_COVERS_DIR.mkdir(parents=True, exist_ok=True)
+
+ALLOWED_TYPES = {"image/png", "image/jpeg", "image/webp"}
+MAX_SIZE = 5 * 1024 * 1024  # 5MB
+
+
+def require_upload_role():
+    return require_role(UserRole.zhipianren, UserRole.zebian)
+
+
+@router.get("/cover")
+def get_current_cover(
+    session: Session = Depends(get_session),
+    current_user: User = Depends(get_current_user),
+):
+    """查询当前题图设置（三角色均可读）。"""
+    row = session.exec(
+        select(CoverSettings).where(CoverSettings.key == "dashboard_cover")
+    ).first()
+    
+    if not row:
+        return {"cover_path": None, "episode_number": None, "episode_title": None}
+    
+    return {
+        "cover_path": row.cover_path,
+        "episode_number": row.episode_number,
+        "episode_title": row.episode_title,
+    }
+
+
+@router.post("/cover")
+def upload_cover(
+    file: UploadFile = File(...),
+    episode_number: int = Form(...),
+    episode_title: str = Form(...),
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_upload_role()),
+):
+    """上传题图（仅制片人/责编）。
+    
+    文件存 static/covers/{uuid}.{ext}，更新 cover_settings 表。
+    """
+    # 校验文件类型
+    if file.content_type not in ALLOWED_TYPES:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail=f"仅支持 PNG/JPG/WEBP，当前：{file.content_type}",
+        )
+
+    # 校验文件大小
+    contents = file.file.read()
+    if len(contents) > MAX_SIZE:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail="文件大小不能超过 5MB",
+        )
+    file.file.seek(0)
+
+    # 生成文件名
+    ext = file.filename.split(".")[-1] if "." in file.filename else "jpg"
+    safe_ext = ext.lower() if ext.lower() in {"png", "jpg", "jpeg", "webp"} else "jpg"
+    filename = f"{uuid.uuid4().hex}.{safe_ext}"
+    filepath = STATIC_COVERS_DIR / filename
+
+    # 写文件
+    with open(filepath, "wb") as f:
+        f.write(contents)
+
+    cover_path = f"{STATIC_COVERS_URL}/{filename}"
+    now = datetime.now()
+    
+    row = session.exec(
+        select(CoverSettings).where(CoverSettings.key == "dashboard_cover")
+    ).first()
+    
+    if row:
+        row.cover_path = cover_path
+        row.episode_number = episode_number
+        row.episode_title = episode_title
+        row.updated_at = now
+    else:
+        row = CoverSettings(
+            key="dashboard_cover",
+            cover_path=cover_path,
+            episode_number=episode_number,
+            episode_title=episode_title,
+            updated_at=now,
+        )
+        session.add(row)
+    
+    session.commit()
+    
+    return {"success": True, "cover_path": cover_path}

backend/app/api/episodes.py

@@ -0,0 +1,103 @@
+"""
+节目期次 API — 增删改查（责编和管理员可操作）
+"""
+
+from fastapi import APIRouter, Depends, HTTPException, status
+from sqlmodel import Session, select
+
+from app.core.deps import get_current_user, require_role
+from app.db.session import get_session
+from app.models.episode import Episode
+from app.models.user import User, UserRole
+from app.schemas.episode import EpisodeCreate, EpisodeResponse, EpisodeUpdate
+
+router = APIRouter(prefix="/api/episodes", tags=["节目期次"])
+
+
+def require_read_role():
+    """GET 接口的鉴权：三个角色都可读"""
+    return require_role(UserRole.zhipianren, UserRole.zebian, UserRole.biandao)
+
+
+@router.get("", response_model=list[EpisodeResponse])
+def list_episodes(
+    limit: int = 50,
+    offset: int = 0,
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_read_role()),
+):
+    """获取节目期次列表（三角色都可读）。按播出日期倒序。"""
+    statement = select(Episode).order_by(Episode.air_date.desc()).offset(offset).limit(limit)
+    episodes = session.exec(statement).all()
+    return episodes
+
+
+@router.get("/{episode_id}", response_model=EpisodeResponse)
+def get_episode(
+    episode_id: int,
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_read_role()),
+):
+    """根据 ID 获取单期节目（三角色都可读）。"""
+    episode = session.get(Episode, episode_id)
+    if episode is None:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="期次不存在")
+    return episode
+
+
+@router.post("", response_model=EpisodeResponse, status_code=status.HTTP_201_CREATED)
+def create_episode(
+    body: EpisodeCreate,
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_role(UserRole.zhipianren, UserRole.zebian)),
+):
+    """新增一期节目（仅管理员/责编）。"""
+    # 检查期次号唯一
+    existing = session.exec(
+        select(Episode).where(Episode.episode_number == body.episode_number)
+    ).first()
+    if existing:
+        raise HTTPException(status_code=status.HTTP_409_CONFLICT, detail="期次号已存在")
+
+    episode = Episode.model_validate(body.model_dump())
+    session.add(episode)
+    session.commit()
+    session.refresh(episode)
+    return episode
+
+
+@router.patch("/{episode_id}", response_model=EpisodeResponse)
+def update_episode(
+    episode_id: int,
+    body: EpisodeUpdate,
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_role(UserRole.zhipianren, UserRole.zebian)),
+):
+    """更新一期节目信息（仅管理员/责编）。"""
+    episode = session.get(Episode, episode_id)
+    if episode is None:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="期次不存在")
+
+    update_data = body.model_dump(exclude_unset=True)
+    for field, value in update_data.items():
+        setattr(episode, field, value)
+
+    session.add(episode)
+    session.commit()
+    session.refresh(episode)
+    return episode
+
+
+@router.delete("/{episode_id}", status_code=status.HTTP_204_NO_CONTENT)
+def delete_episode(
+    episode_id: int,
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_role(UserRole.zhipianren, UserRole.zebian)),
+):
+    """删除一期节目（仅管理员/责编）。"""
+    episode = session.get(Episode, episode_id)
+    if episode is None:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="期次不存在")
+    session.delete(episode)
+    session.commit()
+    return None

backend/app/api/imports.py

@@ -0,0 +1,175 @@
+"""
+Excel 批量导入 API — 模板下载 + 批量导入 + 失败行下载
+"""
+
+import io
+import uuid
+from datetime import date
+
+from fastapi import APIRouter, Depends, HTTPException, Query, UploadFile, File, status
+from fastapi.responses import StreamingResponse
+from openpyxl import Workbook
+from sqlmodel import Session
+
+from app.core.deps import require_role
+from app.db.session import get_session
+from app.models.user import User, UserRole
+from app.schemas.imports import ImportError, ImportResult
+from app.services.excel_service import ExcelService, generate_error_excel
+
+router = APIRouter(prefix="/api/imports", tags=["批量导入"])
+
+
+# ============================================================================
+# 内存失败行存储（进程生命周期，重启后丢失）
+# 仅用于 Phase 2，生产 Phase 4+ 替换为 Redis 或 DB
+# ============================================================================
+_error_store: dict[str, list[dict]] = {}
+
+
+# ============================================================================
+# 模板下载
+# ============================================================================
+
+@router.get("/template")
+def download_template(
+    type: str = Query("episodes", description="模板类型，仅支持 episodes"),
+    current_user: User = Depends(require_role(UserRole.zhipianren, UserRole.zebian)),
+):
+    """下载 episodes 批量导入 Excel 模板。"""
+    if type != "episodes":
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail="Phase 2 不支持排期模板下载",
+        )
+
+    wb = Workbook()
+    ws = wb.active
+    ws.title = "节目期次导入模板"
+
+    # 列说明（第1行）
+    headers = [
+        "episode_number", "program_name", "air_date",
+        "editor_name", "audience_share", "audience_rating",
+        "is_rerun", "notes",
+    ]
+    ws.append(headers)
+
+    # 第2行：示例数据
+    sample = [
+        1, "《军事科技》武器解析系列", "2026-01-15",
+        "张颖", "0.72", "3.5",
+        "否", "首期播出",
+    ]
+    ws.append(sample)
+
+    # 第3行：说明注释（灰色字体）
+    from openpyxl.styles import Font, PatternFill, Alignment
+
+    grey_font = Font(color="808080")
+    note_row = [
+        "期次号（整数，每年从1开始）",
+        "节目名称",
+        "播出日期（YYYY-MM-DD）",
+        "编导姓名（快照，匹配不上则只存姓名）",
+        "收视份额（0-1之间的小数，可空）",
+        "收视率（可空）",
+        "是否重播（是/否），Phase 2 不支持重播导入",
+        "备注",
+    ]
+    ws.append(note_row)
+    for cell in ws[len(headers)]:
+        cell.font = grey_font
+
+    # 第4行：is_rerun 说明
+    rerun_note = ["", "", "", "", "", "", "Phase 2 不支持重播导入，有重播行请空着", ""]
+    ws.append(rerun_note)
+    for cell in ws[4]:
+        cell.font = grey_font
+
+    # 设置列宽
+    ws.column_dimensions["A"].width = 16
+    ws.column_dimensions["B"].width = 28
+    ws.column_dimensions["C"].width = 14
+    ws.column_dimensions["D"].width = 16
+    ws.column_dimensions["E"].width = 16
+    ws.column_dimensions["F"].width = 16
+    ws.column_dimensions["G"].width = 40
+    ws.column_dimensions["H"].width = 20
+
+    output = io.BytesIO()
+    wb.save(output)
+    output.seek(0)
+
+    return StreamingResponse(
+        output,
+        media_type="application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
+        headers={"Content-Disposition": "attachment; filename=episodes_import_template.xlsx"},
+    )
+
+
+# ============================================================================
+# 批量导入
+# ============================================================================
+
+@router.post("/episodes", response_model=ImportResult)
+def import_episodes(
+    file: UploadFile = File(..., description="Excel 文件（.xlsx）"),
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_role(UserRole.zhipianren, UserRole.zebian)),
+):
+    """批量导入节目期次 Excel。"""
+    if not file.filename.endswith(".xlsx"):
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail="仅支持 .xlsx 格式，请使用模板下载获得正确格式",
+        )
+
+    content = file.file.read()
+    service = ExcelService(session)
+
+    try:
+        result = service.import_episodes(content)
+    except ValueError as e:
+        msg = str(e)
+        if "年度目标" in msg:
+            raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail=msg)
+        if "重复" in msg or "库中记录重复" in msg:
+            raise HTTPException(status_code=status.HTTP_409_CONFLICT, detail=msg)
+        # 其他解析/校验错误
+        raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail=msg)
+    except Exception as e:
+        # 捕获 generate_error_excel 等意外错误，避免 500 扩散
+        raise HTTPException(status_code=status.HTTP_500_INTERNAL_SERVER_ERROR, detail=f"导入失败：{str(e)}")
+
+    # 存储失败行（供后续下载）
+    if result["errors"]:
+        _error_store[result["batch_id"]] = result["errors"]
+        result["error_excel_url"] = f"/api/imports/errors/{result['batch_id']}"
+
+    return result
+
+
+# ============================================================================
+# 失败行下载
+# ============================================================================
+
+@router.get("/errors/{batch_id}")
+def download_error_excel(
+    batch_id: str,
+    current_user: User = Depends(require_role(UserRole.zhipianren, UserRole.zebian)),
+):
+    """下载指定批次的失败行 Excel（供责编修正后重新导入）。"""
+    errors = _error_store.get(batch_id)
+    if errors is None:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail="批次不存在或已过期（进程重启后数据丢失），请重新上传",
+        )
+
+    content = generate_error_excel(errors)
+    return StreamingResponse(
+        io.BytesIO(content),
+        media_type="application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
+        headers={"Content-Disposition": f"attachment; filename=import_errors_{batch_id}.xlsx"},
+    )

backend/app/api/knowledge.py

@@ -0,0 +1,132 @@
+"""
+知识库 API — 上传 / 列表 / 删除 / 来源筛选 / 语义搜索
+"""
+
+from typing import Optional
+from pydantic import BaseModel
+
+from fastapi import APIRouter, Depends, HTTPException, UploadFile, File, Query, status
+from sqlmodel import Session
+
+from app.core.deps import get_current_user, require_role
+from app.db.session import get_session
+from app.models.user import User, UserRole
+from app.services.knowledge_service import KnowledgeService
+
+
+router = APIRouter(prefix="/api/knowledge", tags=["知识库"])
+
+
+@router.post("/upload")
+async def upload_md_files(
+    files: list[UploadFile] = File(...),
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_role(UserRole.zhipianren, UserRole.zebian)),
+):
+    """
+    上传单个或多个 .md 文件，解析 frontmatter，写入知识库（含向量）。
+    仅制片人/责编可用。
+    """
+    svc = KnowledgeService()
+    results = []
+    errors = []
+
+    for f in files:
+        if not f.filename.endswith(".md"):
+            errors.append({"file": f.filename, "error": "仅支持 .md 文件"})
+            continue
+        try:
+            content = await f.read()
+            item = svc.store_md_file(content, f.filename)
+            results.append({
+                "id": item.id,
+                "title": item.title,
+                "source_type": item.source_type,
+            })
+        except Exception as e:
+            errors.append({"file": f.filename, "error": str(e)})
+
+    return {"uploaded": results, "errors": errors}
+
+
+@router.get("/items")
+def list_knowledge_items(
+    source_type: Optional[str] = Query(None),
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_role(UserRole.zhipianren, UserRole.zebian, UserRole.biandao)),
+):
+    """
+    知识库列表，支持按 source_type 筛选。三角色均可读。
+    """
+    svc = KnowledgeService()
+    items = svc.list_items(source_type=source_type)
+    return items
+
+
+@router.delete("/items/{item_id}")
+def delete_knowledge_item(
+    item_id: int,
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_role(UserRole.zhipianren, UserRole.zebian)),
+):
+    """
+    删除知识库条目（级联删除向量）。仅制片人/责编可用。
+    """
+    svc = KnowledgeService()
+    deleted = svc.delete_item(item_id)
+    if not deleted:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="条目不存在")
+    return {"message": "删除成功"}
+
+
+@router.get("/sources")
+def list_distinct_sources(
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_role(UserRole.zhipianren, UserRole.zebian, UserRole.biandao)),
+):
+    """
+    返回库里所有不重复的具体出处（source_detail），供来源筛选下拉动态生成。
+    三角色均可读。
+    """
+    svc = KnowledgeService()
+    sources = svc.get_distinct_sources()
+    return [{"source": s} for s in sources]
+
+
+@router.get("/grouped")
+def get_grouped_knowledge_items(
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_role(UserRole.zhipianren, UserRole.zebian, UserRole.biandao)),
+):
+    """
+    返回按 source_type → source_detail 两层聚合的树形结构，
+    含「全部」根节点，供知识库树形导航使用。
+    三角色均可读。
+    """
+    svc = KnowledgeService()
+    return svc.get_grouped_items()
+
+
+class SearchRequest(BaseModel):
+    query: str
+    top_k: int = 5
+
+
+@router.post("/search")
+def search_knowledge(
+    body: SearchRequest,
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_role(UserRole.zhipianren, UserRole.zebian, UserRole.biandao)),
+):
+    """
+    语义检索：输入一段文字，返回最相关的知识库条目及相似度。
+    查询向量用 type="query"（区分于存入时的 type="db"）。
+    三角色均可读。
+    """
+    svc = KnowledgeService()
+    results = svc.search_similar(query_text=body.query, top_k=body.top_k)
+    return {
+        "results": results,
+        "query": body.query,
+        "count": len(results),
+    }

backend/app/api/schedules.py

@@ -0,0 +1,70 @@
+"""
+排期 API — 查询未来排播计划
+
+降级逻辑：
+- schedules 表有数据 → 从 schedules 取未来 limit 条 JOIN episodes
+- schedules 表无数据 → 从 episodes 表取最近 limit 期待播期次（air_date 升序）
+"""
+
+from fastapi import APIRouter, Depends
+from sqlmodel import Session, select
+
+from app.core.deps import get_current_user
+from app.db.session import get_session
+from app.models.user import User
+from app.models.episode import Episode
+from app.models.schedule import Schedule
+
+router = APIRouter(prefix="/api/schedules", tags=["排期"])
+
+
+@router.get("/upcoming")
+def list_upcoming_schedules(
+    limit: int = 6,
+    session: Session = Depends(get_session),
+    current_user: User = Depends(get_current_user),
+):
+    """获取未来排播计划（三角色均可读）。
+    
+    优先从 schedules 表取，schedules 无数据时降级取 episodes 最近待播期次。
+    """
+    # 先查 schedules
+    statement = (
+        select(Schedule, Episode)
+        .join(Episode, Schedule.episode_id == Episode.id, isouter=True)
+        .where(Schedule.planned_air_date >= Episode.air_date)
+        .order_by(Schedule.planned_air_date.asc())
+        .limit(limit)
+    )
+    results = session.exec(statement).all()
+
+    # 降级：从 episodes 直接取（当 schedules 为空或不足时）
+    if not results or len(results) < limit:
+        fallback_stmt = (
+            select(Episode)
+            .order_by(Episode.air_date.asc())
+            .limit(limit)
+        )
+        episodes = session.exec(fallback_stmt).all()
+        return [
+            {
+                "episode_number": ep.episode_number,
+                "program_name": ep.program_name,
+                "planned_air_date": str(ep.air_date),
+                "editor_name_snapshot": ep.editor_name_snapshot,
+            }
+            for ep in episodes
+        ]
+
+    # schedules 有数据
+    output = []
+    for schedule, episode in results:
+        if episode is None:
+            continue
+        output.append({
+            "episode_number": episode.episode_number,
+            "program_name": episode.program_name,
+            "planned_air_date": str(schedule.planned_air_date),
+            "editor_name_snapshot": episode.editor_name_snapshot,
+        })
+    return output

backend/app/api/users.py

@@ -0,0 +1,106 @@
+"""
+用户管理 API — 管理员 CRUD（仅 zhipianren 可操作）
+"""
+
+from fastapi import APIRouter, Depends, HTTPException, status
+from sqlmodel import Session, select
+
+from app.core.deps import get_current_user, require_role
+from app.core.security import hash_password
+from app.db.session import get_session
+from app.models.user import User, UserRole
+from app.schemas.user_admin import UserCreate, UserResponse, UserUpdate
+
+router = APIRouter(prefix="/api/users", tags=["用户管理"])
+
+
+@router.get("", response_model=list[UserResponse])
+def list_users(
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_role(UserRole.zhipianren)),
+):
+    """获取所有用户列表（仅管理员）。"""
+    statement = select(User).order_by(User.id)
+    users = session.exec(statement).all()
+    return users
+
+
+@router.get("/{user_id}", response_model=UserResponse)
+def get_user(
+    user_id: int,
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_role(UserRole.zhipianren)),
+):
+    """根据 ID 获取单个用户（仅管理员）。"""
+    user = session.get(User, user_id)
+    if user is None:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="用户不存在")
+    return user
+
+
+@router.post("", response_model=UserResponse, status_code=status.HTTP_201_CREATED)
+def create_user(
+    body: UserCreate,
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_role(UserRole.zhipianren)),
+):
+    """创建新用户（仅管理员）。"""
+    # 检查用户名唯一
+    existing = session.exec(select(User).where(User.username == body.username)).first()
+    if existing:
+        raise HTTPException(status_code=status.HTTP_409_CONFLICT, detail="用户名已存在")
+
+    user = User(
+        username=body.username,
+        display_name=body.display_name,
+        password_hash=hash_password(body.password),
+        role=body.role,
+        profile_text=body.profile_text,
+    )
+    session.add(user)
+    session.commit()
+    session.refresh(user)
+    return user
+
+
+@router.patch("/{user_id}", response_model=UserResponse)
+def update_user(
+    user_id: int,
+    body: UserUpdate,
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_role(UserRole.zhipianren)),
+):
+    """更新用户信息（仅管理员）。可更新 display_name/role/profile_text/is_active。"""
+    user = session.get(User, user_id)
+    if user is None:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="用户不存在")
+
+    update_data = body.model_dump(exclude_unset=True)
+    for field, value in update_data.items():
+        setattr(user, field, value)
+
+    session.add(user)
+    session.commit()
+    session.refresh(user)
+    return user
+
+
+@router.delete("/{user_id}", status_code=status.HTTP_204_NO_CONTENT)
+def deactivate_user(
+    user_id: int,
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_role(UserRole.zhipianren)),
+):
+    """停用用户（软删除，仅管理员）。"""
+    user = session.get(User, user_id)
+    if user is None:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="用户不存在")
+
+    # 不能停用自己的账号
+    if user.id == current_user.id:
+        raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail="不能停用自己的账号")
+
+    user.is_active = False
+    session.add(user)
+    session.commit()
+    return None

backend/app/api/yearly_targets.py

@@ -0,0 +1,112 @@
+"""
+年度收视目标 API — 增删改查
+
+业务规则（.clinerules 5.2 只增不改）:
+  - yearly_targets.year 有 UNIQUE 约束
+  - INSERT 时 year 冲突 → 直接报错，阻止静默覆盖
+  - 禁止 ON CONFLICT UPDATE
+"""
+
+from fastapi import APIRouter, Depends, HTTPException, status
+from sqlmodel import Session, select
+
+from app.core.deps import get_current_user, require_role
+from app.db.session import get_session
+from app.models.user import User, UserRole
+from app.models.yearly_target import YearlyTarget
+from app.schemas.yearly_target import YearlyTargetCreate, YearlyTargetResponse, YearlyTargetUpdate
+
+router = APIRouter(prefix="/api/yearly_targets", tags=["年度目标"])
+
+
+@router.get("", response_model=list[YearlyTargetResponse])
+def list_targets(
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_role(UserRole.zhipianren, UserRole.zebian, UserRole.biandao)),
+):
+    """获取所有年度目标（三角色均可读）。"""
+    statement = select(YearlyTarget).order_by(YearlyTarget.year.desc())
+    targets = session.exec(statement).all()
+    return targets
+
+
+@router.get("/{target_id}", response_model=YearlyTargetResponse)
+def get_target(
+    target_id: int,
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_role(UserRole.zhipianren, UserRole.zebian, UserRole.biandao)),
+):
+    """根据 ID 获取单个年度目标（三角色均可读）。"""
+    target = session.get(YearlyTarget, target_id)
+    if target is None:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="目标不存在")
+    return target
+
+
+@router.post("", response_model=YearlyTargetResponse, status_code=status.HTTP_201_CREATED)
+def create_target(
+    body: YearlyTargetCreate,
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_role(UserRole.zhipianren, UserRole.zebian)),
+):
+    """新增年度目标（仅管理员/责编）。
+
+    注意：year 有 UNIQUE 约束，重复年份会直接报错。
+    这是预期行为——历史年份目标不可覆盖。
+    """
+    # 先查是否已存在（让报错更友好，而不是数据库层直接崩溃）
+    existing = session.exec(select(YearlyTarget).where(YearlyTarget.year == body.year)).first()
+    if existing:
+        raise HTTPException(
+            status_code=status.HTTP_409_CONFLICT,
+            detail=f"{body.year} 年目标已存在，历史年份不可覆盖。请新增其他年份。",
+        )
+
+    target = YearlyTarget.model_validate(body.model_dump())
+    session.add(target)
+    session.commit()
+    session.refresh(target)
+    return target
+
+
+@router.patch("/{target_id}", response_model=YearlyTargetResponse)
+def update_target(
+    target_id: int,
+    body: YearlyTargetUpdate,
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_role(UserRole.zhipianren, UserRole.zebian)),
+):
+    """更新年度目标（仅管理员/责编）。
+
+    注意：此接口仅用于修正错字等小修，year 字段不可修改。
+    """
+    target = session.get(YearlyTarget, target_id)
+    if target is None:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="目标不存在")
+
+    update_data = body.model_dump(exclude_unset=True)
+    # 禁止修改 year 字段
+    update_data.pop("year", None)
+
+    for field, value in update_data.items():
+        setattr(target, field, value)
+
+    session.add(target)
+    session.commit()
+    session.refresh(target)
+    return target
+
+
+@router.delete("/{target_id}", status_code=status.HTTP_204_NO_CONTENT)
+def delete_target(
+    target_id: int,
+    session: Session = Depends(get_session),
+    current_user: User = Depends(require_role(UserRole.zhipianren)),
+):
+    """删除年度目标（仅制片人）。"""
+    target = session.get(YearlyTarget, target_id)
+    if target is None:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="目标不存在")
+    session.delete(target)
+    session.commit()
+    return None

backend/app/core/config.py

@@ -0,0 +1,36 @@
+"""
+应用配置 — 手动从 .env 读取，不依赖 pydantic-settings 自动 env_file
+"""
+
+import os
+from pathlib import Path
+from dotenv import load_dotenv
+
+# 强制加载 backend/.env 到 os.environ（用 __file__ 定位脚本所在目录，避免 cwd 漂移）
+# __file__ = e:\tps-dashboard\backend\app\core\config.py → parent.parent.parent = e:\tps-dashboard\backend
+_env_path = Path(__file__).parent.parent.parent / ".env"
+load_dotenv(str(_env_path), override=True)
+
+# 直接从 os.environ 读取，手动兜底
+_DATABASE_URL = os.environ.get("DATABASE_URL")
+_SECRET_KEY = os.environ.get("SECRET_KEY", "change-me-to-a-random-string-in-production")
+_SESSION_MAX_AGE = int(os.environ.get("SESSION_MAX_AGE", "86400"))
+
+# MiniMax Embedding API 凭证
+_MINIMAX_EMBED_API_KEY = os.environ.get("MINIMAX_EMBED_API_KEY", "")
+_MINIMAX_GROUP_ID = os.environ.get("MINIMAX_GROUP_ID", "")
+
+# 验证必需配置
+if not _DATABASE_URL:
+    raise RuntimeError(f"[config] DATABASE_URL 未设置。请检查 {_env_path} 是否存在且内容正确。")
+
+# 简单命名空间对象（替代 pydantic-settings，避免 env 读取失效）
+class Settings:
+    DATABASE_URL: str = _DATABASE_URL
+    SECRET_KEY: str = _SECRET_KEY
+    SESSION_MAX_AGE: int = _SESSION_MAX_AGE
+    MINIMAX_EMBED_API_KEY: str = _MINIMAX_EMBED_API_KEY
+    MINIMAX_GROUP_ID: str = _MINIMAX_GROUP_ID
+
+
+settings = Settings()

backend/app/core/deps.py

@@ -0,0 +1,56 @@
+"""
+FastAPI 依赖注入工具 — 获取当前用户、角色鉴权
+"""
+
+from fastapi import Depends, HTTPException, Request, status
+from sqlmodel import Session
+
+from app.db.session import get_session
+from app.models.user import User, UserRole
+
+
+def get_current_user(
+    request: Request,
+    session: Session = Depends(get_session),
+) -> User:
+    """从 Session 中取 user_id，查库返回当前用户。
+
+    未登录或用户不存在 / 已停用时抛出 401。
+    """
+    user_id = request.session.get("user_id")
+    if user_id is None:
+        raise HTTPException(
+            status_code=status.HTTP_401_UNAUTHORIZED,
+            detail="请先登录",
+        )
+    user = session.get(User, user_id)
+    if user is None or not user.is_active:
+        raise HTTPException(
+            status_code=status.HTTP_401_UNAUTHORIZED,
+            detail="用户不存在或已被停用",
+        )
+    return user
+
+
+def require_role(*allowed_roles: UserRole):
+    """返回一个 Depends 依赖，要求当前用户拥有指定的角色之一。
+
+    使用示例:
+        @router.get("/admin-only")
+        def admin_endpoint(user=Depends(require_role(UserRole.zhipianren))):
+            ...
+
+        @router.get("/staff")
+        def staff_endpoint(user=Depends(require_role(UserRole.zhipianren, UserRole.zebian))):
+            ...
+    """
+
+    def check_role(user: User = Depends(get_current_user)) -> User:
+        if user.role not in allowed_roles:
+            raise HTTPException(
+                status_code=status.HTTP_403_FORBIDDEN,
+                detail="权限不足，无法执行此操作",
+            )
+        return user
+
+    return check_role

backend/app/core/security.py

@@ -0,0 +1,17 @@
+"""
+认证安全工具 — passlib + bcrypt
+"""
+
+from passlib.context import CryptContext
+
+pwd_context = CryptContext(schemes=["bcrypt"], deprecated="auto")
+
+
+def hash_password(plain: str) -> str:
+    """用 bcrypt 生成密码哈希。"""
+    return pwd_context.hash(plain)
+
+
+def verify_password(plain: str, hashed: str) -> bool:
+    """验证明文密码与哈希是否匹配。"""
+    return pwd_context.verify(plain, hashed)

backend/app/db/session.py

@@ -0,0 +1,20 @@
+"""
+数据库会话 — SQLModel 同步引擎 + Session 工厂
+"""
+
+from sqlmodel import Session, SQLModel, create_engine
+
+from app.core.config import settings
+
+engine = create_engine(settings.DATABASE_URL, echo=False)
+
+
+def init_db():
+    """创建所有 SQLModel 表（仅开发用，生产由手写 SQL 管理）。"""
+    SQLModel.metadata.create_all(engine)
+
+
+def get_session():
+    """FastAPI 依赖注入: 每个请求获取一个 Session，用完后关闭。"""
+    with Session(engine) as session:
+        yield session

backend/app/main.py

@@ -0,0 +1,58 @@
+"""
+FastAPI 应用入口 — 挂载中间件、路由
+"""
+
+from fastapi import FastAPI
+from starlette.middleware.cors import CORSMiddleware
+from starlette.middleware.sessions import SessionMiddleware
+
+from fastapi.staticfiles import StaticFiles
+from pathlib import Path
+
+from app.api.auth import router as auth_router
+from app.api.episodes import router as episodes_router
+from app.api.imports import router as imports_router
+from app.api.users import router as users_router
+from app.api.yearly_targets import router as yearly_targets_router
+from app.api.dashboard import router as dashboard_router
+from app.api.schedules import router as schedules_router
+from app.api.knowledge import router as knowledge_router
+from app.core.config import settings
+
+app = FastAPI(title="军事科技工作台", version="0.1.0")
+
+# ---- CORS 中间件 ----
+# 显式列 origins 不能用 ['*']，否则 credentials=True 时浏览器拒绝携带 cookie
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["http://localhost:5173"],
+    allow_credentials=True,
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+# ---- Session 中间件 ----
+# 密钥来自 .env，用于签名 session cookie
+app.add_middleware(
+    SessionMiddleware,
+    secret_key=settings.SECRET_KEY,
+    session_cookie="milsci_session",
+    max_age=settings.SESSION_MAX_AGE,
+    same_site="lax",
+    https_only=False,  # 本地开发用 HTTP
+)
+
+# ---- 注册路由 ----
+app.include_router(auth_router)
+app.include_router(episodes_router)
+app.include_router(imports_router)
+app.include_router(yearly_targets_router)
+app.include_router(users_router)
+app.include_router(dashboard_router)
+app.include_router(schedules_router)
+app.include_router(knowledge_router)
+
+# 挂载静态文件目录（题图海报）
+_static_dir = Path(__file__).parent.parent / "static"
+if _static_dir.exists():
+    app.mount("/static", StaticFiles(directory=str(_static_dir)), name="static")

backend/app/models/cover_settings.py

@@ -0,0 +1,28 @@
+"""
+题图设置模型 — SQLModel
+
+只存当前展示的一张题图，不动 episodes 表。
+"""
+
+from datetime import datetime
+
+from sqlalchemy import Column
+from sqlalchemy import DateTime as SADateTime
+from sqlalchemy import Integer as SAInteger
+from sqlalchemy import Text
+from sqlalchemy.sql import func as sa_func
+from sqlmodel import Field, SQLModel
+
+
+class CoverSettings(SQLModel, table=True):
+    __tablename__ = "cover_settings"
+
+    id: int | None = Field(default=None, primary_key=True)
+    key: str = Field(default="dashboard_cover", max_length=50)
+    cover_path: str | None = Field(default=None)        # 当前题图文件路径
+    episode_number: int | None = Field(default=None)   # 关联期次号
+    episode_title: str | None = Field(default=None)    # 关联期次节目名
+    updated_at: datetime | None = Field(
+        default=None,
+        sa_column=Column(SADateTime(timezone=True), nullable=False, server_default=sa_func.now()),
+    )

backend/app/models/episode.py

@@ -0,0 +1,65 @@
+"""
+节目期次模型 — SQLModel
+
+业务规则（.clinerules 5.3 软引用 + 快照模式）:
+  - editor_id 可空（离职编导）
+  - editor_name_snapshot 必填（历史导入也走快照）
+  - 查询历史作品：先 editor_id，回退 editor_name_snapshot 文本匹配
+"""
+
+from datetime import date, datetime
+
+from sqlalchemy import Column
+from sqlalchemy import DateTime as SADateTime
+from sqlalchemy import Text
+from sqlalchemy import Boolean as SABoolean
+from sqlalchemy import Integer as SAInteger
+from sqlalchemy import Numeric as SANumeric
+from sqlalchemy.sql import func as sa_func
+from sqlmodel import Field, SQLModel
+
+
+class Episode(SQLModel, table=True):
+    __tablename__ = "episodes"
+
+    id: int | None = Field(default=None, primary_key=True)
+    episode_number: int = Field(unique=True, index=True)
+    program_name: str = Field(max_length=200)
+    air_date: date
+
+    # 编导（软引用 + 快照双字段）
+    # 在职编导：editor_id 和 editor_name_snapshot 都填
+    # 离职编导：editor_id 留空，editor_name_snapshot 填姓名
+    editor_id: int | None = Field(default=None, foreign_key="users.id")
+    editor_name_snapshot: str = Field(max_length=100)
+
+    # 收视数据
+    # 注意：.clinerules 5.1 颜色只对 audience_share 上色
+    # audience_rating 只展示数字，不参与颜色判定
+    audience_share: float | None = Field(default=None)
+    audience_rating: float | None = Field(default=None)
+
+    # 重播相关
+    is_rerun: bool = Field(default=False)
+    original_episode_id: int | None = Field(default=None, foreign_key="episodes.id")
+    is_time_sensitive: bool = Field(default=False)
+
+    # 外拍信息
+    is_outdoor_shoot: bool = Field(default=False)
+    outdoor_contact_person: str | None = Field(default=None, max_length=100)
+    outdoor_contact_info: str | None = Field(default=None, max_length=200)
+
+    # 线索
+    lead_source: str | None = Field(default=None, max_length=300)
+
+    # 责编备注
+    notes: str | None = Field(default=None)
+
+    created_at: datetime | None = Field(
+        default=None,
+        sa_column=Column(SADateTime(timezone=True), nullable=False, server_default=sa_func.now()),
+    )
+    updated_at: datetime | None = Field(
+        default=None,
+        sa_column=Column(SADateTime(timezone=True), nullable=False, server_default=sa_func.now()),
+    )

backend/app/models/knowledge.py

@@ -0,0 +1,55 @@
+"""
+知识库模型 — SQLModel
+对应 knowledge_items 和 knowledge_embeddings 两张表
+embedding 字段使用 pgvector.Vector（对应 PG vector(1536)）
+"""
+
+from datetime import datetime, date
+from typing import Optional, Any
+
+from sqlalchemy import Column, DateTime as SADateTime, Text, Integer
+from sqlalchemy.dialects.postgresql import JSONB
+from sqlalchemy.sql import func as sa_func
+from sqlmodel import Field, SQLModel
+
+from pgvector.sqlalchemy import Vector
+
+
+class KnowledgeItem(SQLModel, table=True):
+    """知识库条目（knowledge_items）"""
+    __tablename__ = "knowledge_items"
+
+    id: Optional[int] = Field(default=None, primary_key=True)
+    title: str = Field(max_length=300)
+    content_md: Optional[str] = Field(default=None)
+    source_type: str = Field(default="manual", max_length=30)
+    source_file_name: Optional[str] = Field(default=None, max_length=300)
+    source_url: Optional[str] = Field(default=None, max_length=1000)
+    author: Optional[str] = Field(default=None, max_length=100)
+    publish_date: Optional[date] = Field(default=None)
+    tags: Any = Field(default=None, sa_column=Column(JSONB, default=[]))
+    related_entities: Any = Field(default=None, sa_column=Column(JSONB, default=[]))
+    related_concepts: Any = Field(default=None, sa_column=Column(JSONB, default=[]))
+    created_at: datetime | None = Field(
+        default=None,
+        sa_column=Column(SADateTime(timezone=True), nullable=False, server_default=sa_func.now()),
+    )
+    updated_at: datetime | None = Field(
+        default=None,
+        sa_column=Column(SADateTime(timezone=True), nullable=False, server_default=sa_func.now()),
+    )
+
+
+class KnowledgeEmbedding(SQLModel, table=True):
+    """知识库向量（knowledge_embeddings）"""
+    __tablename__ = "knowledge_embeddings"
+
+    id: Optional[int] = Field(default=None, primary_key=True)
+    knowledge_id: int = Field(foreign_key="knowledge_items.id", index=True)
+    chunk_index: int = Field(default=0)
+    chunk_text: str = Field(sa_column=Column(Text, nullable=False))
+    embedding: Any = Field(sa_column=Column(Vector(1536), nullable=False))
+    created_at: datetime | None = Field(
+        default=None,
+        sa_column=Column(SADateTime(timezone=True), nullable=False, server_default=sa_func.now()),
+    )

backend/app/models/schedule.py

@@ -0,0 +1,33 @@
+"""
+排期模型 — SQLModel
+"""
+
+from datetime import date, datetime
+
+from sqlalchemy import Column
+from sqlalchemy import DateTime as SADateTime
+from sqlalchemy import Integer as SAInteger
+from sqlalchemy import Date as SADate
+from sqlalchemy import Text
+from sqlalchemy.sql import func as sa_func
+from sqlmodel import Field, SQLModel
+
+
+class Schedule(SQLModel, table=True):
+    __tablename__ = "schedules"
+
+    id: int | None = Field(default=None, primary_key=True)
+    episode_id: int | None = Field(default=None, foreign_key="episodes.id")
+    planned_air_date: date
+    status: str = Field(default="planned", max_length=20)
+    editor_id: int | None = Field(default=None, foreign_key="users.id")
+    notes: str | None = Field(default=None)
+
+    created_at: datetime | None = Field(
+        default=None,
+        sa_column=Column(SADateTime(timezone=True), nullable=False, server_default=sa_func.now()),
+    )
+    updated_at: datetime | None = Field(
+        default=None,
+        sa_column=Column(SADateTime(timezone=True), nullable=False, server_default=sa_func.now()),
+    )

backend/app/models/user.py

@@ -0,0 +1,44 @@
+"""
+用户模型 — SQLModel
+
+角色体系（不再改）:
+  - zhipianren = 制片人 / 管理员 = 全部权限
+  - zebian     = 责编 = 维护仪表盘 / 知识库 / 排期
+  - biandao    = 编导 = 使用 TPS / 查看知识库 / 生成报题单 / 个人热点雷达
+"""
+
+from enum import Enum
+from datetime import datetime, timezone
+
+from sqlalchemy import Column
+from sqlalchemy import DateTime as SADateTime
+from sqlalchemy.sql import func as sa_func
+from sqlmodel import Field, SQLModel
+
+
+class UserRole(str, Enum):
+    """用户角色枚举，值与数据库 CHECK 约束一致。"""
+    zhipianren = "zhipianren"   # 制片人 / 管理员
+    zebian = "zebian"           # 责编
+    biandao = "biandao"         # 编导
+
+
+class User(SQLModel, table=True):
+    __tablename__ = "users"
+
+    id: int | None = Field(default=None, primary_key=True)
+    username: str = Field(max_length=50, unique=True, index=True)
+    display_name: str = Field(max_length=100)
+    password_hash: str = Field(max_length=128)
+    role: UserRole = Field(default=UserRole.biandao, max_length=20)
+    profile_text: str | None = Field(default=None)
+    is_active: bool = Field(default=True)
+
+    created_at: datetime | None = Field(
+        default=None,
+        sa_column=Column(SADateTime(timezone=True), nullable=False, server_default=sa_func.now()),
+    )
+    updated_at: datetime | None = Field(
+        default=None,
+        sa_column=Column(SADateTime(timezone=True), nullable=False, server_default=sa_func.now()),
+    )

backend/app/models/yearly_target.py

@@ -0,0 +1,34 @@
+"""
+年度收视目标模型 — SQLModel
+
+业务规则（.clinerules 5.2 年度目标只增不改）:
+  - yearly_targets.year 有 UNIQUE 约束，数据库层已保证
+  - 代码端 INSERT 时只许 ON CONFLICT DO NOTHING，禁止 ON CONFLICT UPDATE
+  - 历史年份目标行绝不可被覆盖或删除
+  - 前端判颜色时按播出年份取对应年的目标，不是当前年份
+"""
+
+from datetime import date, datetime
+
+from sqlalchemy import Column
+from sqlalchemy import DateTime as SADateTime
+from sqlalchemy import Integer as SAInteger
+from sqlalchemy import Numeric as SANumeric
+from sqlalchemy.sql import func as sa_func
+from sqlmodel import Field, SQLModel
+
+
+class YearlyTarget(SQLModel, table=True):
+    __tablename__ = "yearly_targets"
+
+    id: int | None = Field(default=None, primary_key=True)
+    year: int = Field(unique=True, index=True)
+    base_target: float  # 基础目标
+    stretch_target: float  # 摸高目标
+    issued_date: date | None = Field(default=None)
+    notes: str | None = Field(default=None)
+
+    created_at: datetime | None = Field(
+        default=None,
+        sa_column=Column(SADateTime(timezone=True), nullable=False, server_default=sa_func.now()),
+    )

backend/app/schemas/auth.py

@@ -0,0 +1,24 @@
+"""
+认证相关 Pydantic Schema — 请求 / 响应模型
+"""
+
+from pydantic import BaseModel, ConfigDict
+
+from app.models.user import UserRole
+
+
+class LoginRequest(BaseModel):
+    username: str
+    password: str
+
+
+class UserResponse(BaseModel):
+    """返回给前端的用户信息（不含 password_hash）。"""
+    id: int
+    username: str
+    display_name: str
+    role: UserRole
+    profile_text: str | None = None
+    is_active: bool = True
+
+    model_config = ConfigDict(from_attributes=True)

backend/app/schemas/episode.py

@@ -0,0 +1,41 @@
+"""
+节目期次 Schema — 请求 / 响应模型
+"""
+
+from datetime import date
+
+from pydantic import BaseModel, ConfigDict
+
+from app.models.user import UserRole
+
+
+class EpisodeBase(BaseModel):
+    episode_number: int
+    program_name: str
+    air_date: date
+    editor_id: int | None = None
+    editor_name_snapshot: str
+    audience_share: float | None = None
+    audience_rating: float | None = None
+    is_rerun: bool = False
+    original_episode_id: int | None = None
+    is_time_sensitive: bool = False
+    is_outdoor_shoot: bool = False
+    outdoor_contact_person: str | None = None
+    outdoor_contact_info: str | None = None
+    lead_source: str | None = None
+    notes: str | None = None
+
+
+class EpisodeCreate(EpisodeBase):
+    pass
+
+
+class EpisodeUpdate(EpisodeBase):
+    pass
+
+
+class EpisodeResponse(EpisodeBase):
+    id: int
+
+    model_config = ConfigDict(from_attributes=True)

backend/app/schemas/imports.py

@@ -0,0 +1,28 @@
+"""
+导入相关 Schema — 请求 / 响应模型
+"""
+
+from datetime import date
+from typing import Optional
+from pydantic import BaseModel, ConfigDict
+
+
+class ImportError(BaseModel):
+    """单行失败详情"""
+    row_number: int          # Excel 行号（第2行=第1条数据）
+    reason: str              # 失败原因
+    raw_data: dict           # 原始行数据（字典形式）
+
+    model_config = ConfigDict(from_attributes=True)
+
+
+class ImportResult(BaseModel):
+    """批量导入响应"""
+    batch_id: str            # UUID，用于下载失败行 Excel
+    total_rows: int           # 总行数
+    success_count: int        # 成功数
+    failed_count: int        # 失败数
+    errors: list[ImportError]  # 失败行列表
+    error_excel_url: Optional[str] = None  # 失败行 Excel 下载 URL（无可失败行时可不返回）
+
+    model_config = ConfigDict(from_attributes=True)

backend/app/schemas/user_admin.py

@@ -0,0 +1,33 @@
+"""
+用户管理 Schema — 管理员 CRUD（仅 zhipianren 可用）
+"""
+
+from pydantic import BaseModel, ConfigDict
+
+from app.models.user import UserRole
+
+
+class UserCreate(BaseModel):
+    username: str
+    display_name: str
+    password: str
+    role: UserRole
+    profile_text: str | None = None
+
+
+class UserUpdate(BaseModel):
+    display_name: str | None = None
+    role: UserRole | None = None
+    profile_text: str | None = None
+    is_active: bool | None = None
+
+
+class UserResponse(BaseModel):
+    id: int
+    username: str
+    display_name: str
+    role: UserRole
+    profile_text: str | None = None
+    is_active: bool = True
+
+    model_config = ConfigDict(from_attributes=True)

backend/app/schemas/yearly_target.py

@@ -0,0 +1,36 @@
+"""
+年度收视目标 Schema — 请求 / 响应模型
+
+业务规则：yearly_targets 只增不改，ON CONFLICT DO NOTHING
+"""
+
+from datetime import date
+
+from pydantic import BaseModel, ConfigDict
+
+
+class YearlyTargetBase(BaseModel):
+    year: int
+    base_target: float
+    stretch_target: float
+    issued_date: date | None = None
+    notes: str | None = None
+
+
+class YearlyTargetCreate(YearlyTargetBase):
+    pass
+
+
+class YearlyTargetUpdate(BaseModel):
+    """PATCH 时允许只传部分字段，year 不可修改（由 API 层 pop 掉）。"""
+    year: int | None = None
+    base_target: float | None = None
+    stretch_target: float | None = None
+    issued_date: date | None = None
+    notes: str | None = None
+
+
+class YearlyTargetResponse(YearlyTargetBase):
+    id: int
+
+    model_config = ConfigDict(from_attributes=True)

backend/app/services/embedding_service.py

@@ -0,0 +1,69 @@
+"""
+Embedding 调用服务 — 封装 MiniMax embo-01
+
+请求格式（确认自探路脚本）：
+  POST /v1/embeddings
+  Body: {"model": "embo-01", "texts": [...], "type": "db"|"query"}
+响应格式：
+  {"vectors": [[...1536 floats...]], "total_tokens": N, "base_resp": {"status_code": 0, "status_msg": "success"}}
+"""
+
+import httpx
+from typing import List
+
+from app.core.config import settings
+
+
+class EmbeddingService:
+    """MiniMax embo-01 embedding 调用封装"""
+
+    def __init__(self):
+        self.api_key = settings.MINIMAX_EMBED_API_KEY
+        self.group_id = settings.MINIMAX_GROUP_ID
+        self.endpoint = "https://api.minimax.chat/v1/embeddings"
+
+    def embed(self, texts: List[str], embed_type: str = "db") -> List[List[float]]:
+        """
+        调用 embo-01 将文本列表转为向量
+        
+        Args:
+            texts: 文本列表（支持批量）
+            embed_type: "db" = 存入库，"query" = 查询
+        Returns:
+            List[List[float]]，每个元素是一组 1536 维向量
+        """
+        if not self.api_key or self.api_key == "your_api_key_here":
+            raise RuntimeError("MINIMAX_EMBED_API_KEY not configured in .env")
+        if not self.group_id or self.group_id == "your_group_id_here":
+            raise RuntimeError("MINIMAX_GROUP_ID not configured in .env")
+
+        headers = {
+            "Authorization": f"Bearer {self.api_key}",
+            "GroupId": self.group_id,
+            "Content-Type": "application/json",
+        }
+        payload = {
+            "model": "embo-01",
+            "texts": texts,
+            "type": embed_type,
+        }
+
+        resp = httpx.post(self.endpoint, headers=headers, json=payload, timeout=60.0)
+        resp.raise_for_status()
+        data = resp.json()
+
+        # 检查业务错误
+        base_resp = data.get("base_resp", {})
+        if base_resp.get("status_code", 0) != 0:
+            raise RuntimeError(f"Embedding API error: {base_resp.get('status_msg', 'unknown')}")
+
+        vectors = data.get("vectors", [])
+        if not vectors:
+            raise RuntimeError("No vectors returned from embedding API")
+
+        return vectors
+
+    def embed_single(self, text: str, embed_type: str = "db") -> List[float]:
+        """单文本 embedding，返回 1536 维向量列表（Python list）"""
+        vectors = self.embed([text], embed_type=embed_type)
+        return vectors[0]

backend/app/services/excel_service.py

@@ -0,0 +1,247 @@
+"""
+Excel 解析服务 — 批量导入核心逻辑
+
+业务规则：
+  - 前置校验：所有涉及年份的 yearly_targets 必须存在，否则整体报错（不入库任何行）
+  - 幂等：检测到重复 (year, episode_number) → 整体 409，不入库
+  - 事务：逐行提交，成功立即 commit，失败行收集到 errors[]
+  - 软引用+快照：编导姓名匹配不到在职用户 → editor_id=NULL + editor_name_snapshot=姓名
+  - is_rerun 解析：接受是/否、true/false、1/0，大小写不敏感
+  - Phase 2 不支持重播行（is_rerun=是 → 报错标记为失败行）
+"""
+
+import io
+import uuid
+from datetime import date
+from typing import Any
+
+import pandas as pd
+from sqlmodel import Session, select
+
+from app.models.episode import Episode
+from app.models.user import User
+from app.models.yearly_target import YearlyTarget
+
+
+TRUTHY = {"是", "true", "1", "yes"}
+FALSY = {"否", "false", "0", "no"}
+
+
+def parse_bool(val: Any) -> bool | None:
+    """解析布尔值，接受多种格式。"""
+    if val is None:
+        return None
+    v = str(val).strip().lower()
+    if v == "":
+        return None
+    if v in TRUTHY:
+        return True
+    if v in FALSY:
+        return False
+    raise ValueError(f"无法解析布尔值: {val!r}")
+
+
+def parse_date(val: Any) -> date | None:
+    """解析日期，支持 YYYY-MM-DD 和 YYYY/MM/DD。"""
+    if val is None or str(val).strip() == "":
+        return None
+    v = str(val).strip()
+    for fmt in ("%Y-%m-%d", "%Y/%m/%d", "%Y%m%d"):
+        try:
+            return date.fromisoformat(v)
+        except ValueError:
+            pass
+    raise ValueError(f"无法解析日期: {val!r}")
+
+
+def parse_float(val: Any) -> float | None:
+    """解析浮点数，接受百分比字符串。"""
+    if val is None:
+        return None
+    v = str(val).strip()
+    if v == "":
+        return None
+    if v.endswith("%"):
+        v = v[:-1]
+    return float(v)
+
+
+def find_editor_by_name(session: Session, name: str):
+    """按 display_name 匹配在职编导。匹配上返 (id, name)，匹配不上返 (None, name)。"""
+    if not name or str(name).strip() == "":
+        return None, ""
+    name = str(name).strip()
+    user = session.exec(
+        select(User).where(User.display_name == name, User.is_active == True)
+    ).first()
+    if user:
+        return user.id, name
+    return None, name
+
+
+class ExcelService:
+    def __init__(self, session: Session):
+        self.session = session
+        self.errors: list[dict] = []
+        self.batch_id = str(uuid.uuid4())
+
+    def validate_yearly_targets(self, air_dates: list[date]):
+        """前置校验：所有涉及年份的 yearly_targets 必须存在，否则整体报错。"""
+        distinct_years = set(dt.year for dt in air_dates if dt is not None)
+        missing = []
+        for yr in sorted(distinct_years):
+            exists = self.session.exec(
+                select(YearlyTarget).where(YearlyTarget.year == yr)
+            ).first()
+            if not exists:
+                missing.append(yr)
+        if missing:
+            raise ValueError(
+                f"导入前请先录入以下年份的年度目标：{', '.join(map(str, missing))}。"
+                "请先在年度目标页录入目标后再导入。"
+            )
+
+    def check_duplicates(self, rows: list[dict]):
+        """检测重复 (year, episode_number)，有重复则整体报错。"""
+        seen = set()
+        duplicates = []
+        for row in rows:
+            key = (row["_air_year"], row["episode_number"])
+            if key in seen:
+                duplicates.append(row)
+            seen.add(key)
+        if duplicates:
+            dup_list = [
+                f"{r['_air_year']}年第 {r['episode_number']} 期"
+                for r in duplicates
+            ]
+            raise ValueError(
+                f"文件中以下期次与库中记录重复：{', '.join(dup_list)}。"
+                "请先手动删除重复期次后再重新导入。"
+            )
+
+    def import_episodes(self, file_content: bytes) -> dict:
+        """解析 Excel，批量导入 episodes。返回 ImportResult 结构。"""
+        # 1. 读取 Excel（pandas 2.x 要求 file-like object，用 BytesIO 包装 bytes）
+        file_like = io.BytesIO(file_content)
+        df = pd.read_excel(file_like, engine="openpyxl")
+        rows = df.to_dict(orient="records")
+
+        # 2. 解析 air_date 并提取年份
+        parsed_rows = []
+        air_dates = []
+        for i, row in enumerate(rows, start=2):  # Excel 行号从2开始（第1行=表头）
+            raw = dict(row)
+            try:
+                air_dt = parse_date(row.get("air_date"))
+                if air_dt is None:
+                    raise ValueError("air_date 不能为空")
+                parsed_rows.append({
+                    "_row_number": i,
+                    "_air_year": air_dt.year,
+                    "_air_date": air_dt,
+                    "episode_number": int(row["episode_number"]),
+                    "program_name": str(row["program_name"]).strip(),
+                    "audience_share": parse_float(row.get("audience_share")),
+                    "audience_rating": parse_float(row.get("audience_rating")),
+                    "is_rerun": parse_bool(row.get("is_rerun")),
+                    "editor_name_snapshot": str(row.get("editor_name", "")).strip() or "未知编导",
+                    "notes": str(row.get("notes", "")).strip() or None,
+                    "_raw": raw,
+                })
+                air_dates.append(air_dt)
+            except Exception as e:
+                self.errors.append({
+                    "row_number": i,
+                    "reason": str(e),
+                    "raw_data": raw,
+                })
+
+        # 3. 前置校验：年份 targets
+        try:
+            self.validate_yearly_targets(air_dates)
+        except ValueError:
+            raise  # 直接抛给调用方，整体 400
+
+        # 4. 重复检测
+        try:
+            self.check_duplicates(parsed_rows)
+        except ValueError:
+            raise  # 直接抛给调用方，整体 409
+
+        # 5. 逐行入库
+        for row in parsed_rows:
+            try:
+                self._import_one_row(row)
+            except Exception as e:
+                self.errors.append({
+                    "row_number": row["_row_number"],
+                    "reason": str(e),
+                    "raw_data": row["_raw"],
+                })
+
+        # 6. 计算结果
+        total = len(rows)
+        success = total - len(self.errors)
+        return {
+            "batch_id": self.batch_id,
+            "total_rows": total,
+            "success_count": success,
+            "failed_count": len(self.errors),
+            "errors": self.errors,
+        }
+
+    def _import_one_row(self, row: dict):
+        """导入单行，处理 is_rerun 逻辑。"""
+        # 重播行 Phase 2 不支持
+        if row["is_rerun"] is True:
+            raise ValueError(
+                "Phase 2 不支持重播期次导入。"
+                "有重播行请空着或用其他工具录入，系统已标记为失败行。"
+            )
+
+        # 编辑匹配
+        editor_id, editor_name = find_editor_by_name(self.session, row["editor_name_snapshot"])
+        if editor_id is None and row["editor_name_snapshot"] == "未知编导":
+            editor_name = row["editor_name_snapshot"]
+
+        # 插入
+        episode = Episode(
+            episode_number=row["episode_number"],
+            program_name=row["program_name"],
+            air_date=row["_air_date"],
+            editor_id=editor_id,
+            editor_name_snapshot=editor_name,
+            audience_share=row["audience_share"],
+            audience_rating=row["audience_rating"],
+            is_rerun=False,
+            original_episode_id=None,
+            notes=row["notes"],
+        )
+        self.session.add(episode)
+        self.session.commit()
+        self.session.refresh(episode)
+
+
+def generate_error_excel(errors: list[dict]) -> bytes:
+    """生成失败行 Excel，供责编下载修正。"""
+    import io
+    from openpyxl import Workbook
+
+    wb = Workbook()
+    ws = wb.active
+    ws.title = "失败行"
+
+    # 表头
+    headers = ["row_number", "reason"] + list(errors[0]["raw_data"].keys()) if errors else ["row_number", "reason"]
+    ws.append(headers)
+
+    # 失败行
+    for err in errors:
+        row_data = [err["row_number"], err["reason"]] + list(err["raw_data"].values())
+        ws.append(row_data)
+
+    output = io.BytesIO()
+    wb.save(output)
+    output.seek(0)
+    return output.read()

backend/app/services/knowledge_service.py

@@ -0,0 +1,354 @@
+"""
+知识库服务 — 写入向量 + 语义检索 + md 文件解析
+使用 pgvector 原生 SQL 向量检索（<=> 余弦距离算子），不在 Python 侧计算
+"""
+
+from typing import Optional
+from datetime import date
+
+import frontmatter
+from sqlalchemy import text
+from sqlmodel import Session, select
+from pgvector.sqlalchemy import Vector
+
+from app.models.knowledge import KnowledgeItem, KnowledgeEmbedding
+from app.services.embedding_service import EmbeddingService
+from app.db.session import engine
+
+
+class KnowledgeService:
+    """知识库 CRUD + 语义检索 + md 解析"""
+
+    # yaml 类型字段 → source_type 枚举映射
+    SOURCE_TYPE_MAP = {
+        "杂志文章": "military_report",
+        "军报": "military_report",
+        "节目文稿": "manuscript",
+        "报题单": "baoti",
+    }
+
+    # 来源大类固定显示顺序（制片人 Obsidian 习惯）
+    SOURCE_TYPE_ORDER = [
+        "manuscript",      # 节目文稿
+        "military_report", # 杂志文章
+        "baoti",           # 报题单
+        "manual",          # 其他
+    ]
+
+    # 二级分组维度映射（与前端 useKnowledgeGrouping.js 的 SECONDARY_GROUP_FIELD 一致）
+    # key = source_type, value = 用来做二级分组的字段名，None = 不建二级节点
+    SECONDARY_GROUP_FIELD = {
+        "manuscript": "author",              # 节目文稿 → 按作者(编导)
+        "military_report": "source_detail",  # 杂志文章 → 按出处
+        "baoti": None,                       # 报题单 → 不分组
+        "manual": None,                      # 其他 → 不分组
+    }
+
+    SOURCE_TYPE_LABEL = {
+        "military_report": "杂志文章",
+        "manuscript": "节目文稿",
+        "baoti": "报题单",
+        "manual": "其他",
+    }
+
+    def __init__(self):
+        self.embedder = EmbeddingService()
+
+    def parse_md_file(self, file_content: bytes, file_name: str) -> dict:
+        """
+        解析一个 .md 文件的 yaml frontmatter + 正文，返回入库用的字典。
+        严格按真实样本的字段名映射，不猜测。
+
+        Returns:
+            dict 含 keys: title, content_md, source_type, author, publish_date,
+            source_detail, metadata(JSONB), related_entities(JSONB)
+        """
+        content = file_content.decode("utf-8", errors="replace")
+        parsed = frontmatter.loads(content)
+        fm = parsed.metadata or {}
+
+        # —— 类型 → source_type（硬映射，不猜测）——
+        raw_type = str(fm.get("类型", "")).strip()
+        source_type = self.SOURCE_TYPE_MAP.get(raw_type, "manual")
+
+        # —— 标题：名称 或 标题——
+        title = str(fm.get("名称", "") or fm.get("标题", "")).strip()
+        if not title:
+            # fallback: 用正文第一行或文件名
+            lines = [l.strip() for l in content.split("\n") if l.strip() and not l.strip().startswith("---")]
+            title = lines[0] if lines else file_name
+
+        # —— 作者：作者 或 编导——
+        author = str(fm.get("作者", "") or fm.get("编导", "") or "").strip() or None
+
+        # —— 出处详情：期刊 + 期号（拼在一起存进 JSONB 的 source_detail）——
+        journal = str(fm.get("期刊", "") or "").strip()
+        issue = str(fm.get("期号", "") or "").strip()
+        if journal or issue:
+            source_detail = f"{journal} {issue}".strip()
+        else:
+            source_detail = None
+
+        # —— 播出日期：容错 "待补充" 等非日期文本——
+        raw_date = str(fm.get("播出日期", "") or "").strip()
+        publish_date = None
+        if raw_date and raw_date not in ("待补充", "待确认", ""):
+            try:
+                publish_date = date.fromisoformat(raw_date)
+            except ValueError:
+                # 非 ISO 格式，尝试 common 格式
+                for fmt in ("%Y-%m-%d", "%Y年%m月%d日", "%Y/%m/%d"):
+                    try:
+                        publish_date = date.fromisoformat(raw_date.replace("年", "-").replace("月", "-").replace("日", ""))
+                        break
+                    except ValueError:
+                        continue
+
+        # —— 权重（不展示，存 JSONB 备 Phase 4）——
+        weight = str(fm.get("权重", "") or "").strip() or None
+
+        # —— 相关实体（涉及装备/涉及技术/涉及厂商/主题）——
+        related_entities = []
+        for key in ("涉及装备", "涉及技术", "涉及厂商", "主题"):
+            val = fm.get(key)
+            if val:
+                if isinstance(val, list):
+                    related_entities.extend(val)
+                elif isinstance(val, str):
+                    # 可能是 "山东舰, 福建舰" 这样的逗号分隔字符串
+                    for item in val.replace("，", ",").split(","):
+                        item = item.strip()
+                        if item:
+                            related_entities.append(item)
+
+        # —— metadata JSONB：权重、出处详情、双链预留——
+        metadata = {}
+        if weight:
+            metadata["weight"] = weight
+        if source_detail:
+            metadata["source_detail"] = source_detail
+        # related_concepts 字段预留给双链解析（Phase 4），本 Task 原样存入
+        metadata["double_bracket_links"] = self._extract_double_brackets(parsed.content)
+
+        # —— 正文（去掉 frontmatter 的纯内容）——
+        content_md = parsed.content
+
+        return {
+            "title": title,
+            "content_md": content_md,
+            "source_type": source_type,
+            "author": author,
+            "publish_date": publish_date,
+            "metadata": metadata if metadata else None,
+            "related_entities": related_entities if related_entities else None,
+            "source_file_name": file_name,
+        }
+
+    def _extract_double_brackets(self, text: str) -> list[str]:
+        """提取 [[...]] 双链标记，原样返回列表，不解析成图谱（本 Task 留门）。"""
+        import re
+        return re.findall(r"\[\[([^\]]+)\]\]", text)
+
+    def store_md_file(self, file_content: bytes, file_name: str) -> KnowledgeItem:
+        """
+        读取一篇 md 内容，调用 embo-01 拿到向量，写入 knowledge_items + knowledge_embeddings
+        """
+        parsed = self.parse_md_file(file_content, file_name)
+
+        # 调用 embedding（type="db" 表示存入知识库）
+        embedding_list = self.embedder.embed_single(parsed["content_md"], embed_type="db")
+
+        with Session(engine) as session:
+            item = KnowledgeItem(
+                title=parsed["title"],
+                content_md=parsed["content_md"],
+                source_type=parsed["source_type"],
+                source_file_name=parsed["source_file_name"],
+                author=parsed["author"],
+                publish_date=parsed["publish_date"],
+                tags=parsed["metadata"],
+                related_entities=parsed["related_entities"],
+            )
+            session.add(item)
+            session.flush()
+
+            emb = KnowledgeEmbedding(
+                knowledge_id=item.id,
+                chunk_index=0,
+                chunk_text=parsed["content_md"],
+                embedding=embedding_list,
+            )
+            session.add(emb)
+            session.commit()
+            session.refresh(item)
+            return item
+
+    def delete_item(self, knowledge_id: int) -> bool:
+        """删除知识库条目及其向量（CASCADE 已由 DB 层配置）。"""
+        with Session(engine) as session:
+            item = session.get(KnowledgeItem, knowledge_id)
+            if item is None:
+                return False
+            session.delete(item)
+            session.commit()
+            return True
+
+    def list_items(self, source_type: Optional[str] = None) -> list[dict]:
+        """返回知识库条目列表（含 source_detail 从 metadata 解压）。"""
+        with Session(engine) as session:
+            statement = select(KnowledgeItem).order_by(KnowledgeItem.created_at.desc())
+            if source_type:
+                statement = statement.where(KnowledgeItem.source_type == source_type)
+            items = session.exec(statement).all()
+
+            results = []
+            for item in items:
+                # 从 tags(JSONB) 取 source_detail
+                tags = item.tags or {}
+                source_detail = tags.get("source_detail") if isinstance(tags, dict) else None
+                results.append({
+                    "id": item.id,
+                    "title": item.title,
+                    "author": item.author,
+                    "publish_date": item.publish_date,
+                    "source_type": item.source_type,
+                    "source_file_name": item.source_file_name,
+                    "source_detail": source_detail,
+                    "created_at": item.created_at,
+                })
+            return results
+
+    def get_distinct_sources(self) -> list[str]:
+        """返回库里所有不重复的 source_detail（动态从 JSONB 提取），供筛选下拉用。"""
+        with Session(engine) as session:
+            items = session.exec(select(KnowledgeItem)).all()
+            sources = set()
+            for item in items:
+                tags = item.tags or {}
+                if isinstance(tags, dict) and tags.get("source_detail"):
+                    sources.add(tags["source_detail"])
+            return sorted(list(sources))
+
+    def search_similar(self, query_text: str, top_k: int = 5) -> list[dict]:
+        """
+        语义检索：查询句转为向量，用 SQL 余弦距离（<=>）在数据库层检索
+        返回 top_k 条相似笔记，含相似度分数 + 原文片段（SQL 端截断前 200 字）。
+
+        注意：当前取前 200 字是已知妥协（整篇向量检索无法定位中段命中点），
+        Phase 4a 做切块检索（chunk）时可优化为取最相关片段。
+        """
+        query_vector = self.embedder.embed_single(query_text, embed_type="query")
+        vec_str = "[" + ",".join(str(v) for v in query_vector) + "]"
+
+        with Session(engine) as session:
+            sql = f"""
+                SELECT
+                    ki.id,
+                    ki.title,
+                    ki.source_type,
+                    ki.author,
+                    ki.tags,
+                    SUBSTRING(ki.content_md, 1, 200) AS snippet,
+                    1 - (ke.embedding <=> '{vec_str}'::vector) AS similarity
+                FROM knowledge_embeddings ke
+                JOIN knowledge_items ki ON ke.knowledge_id = ki.id
+                WHERE ke.chunk_index = 0
+                ORDER BY ke.embedding <=> '{vec_str}'::vector
+                LIMIT {top_k}
+            """
+            stmt = text(sql)
+            rows = session.execute(stmt).all()
+            results = []
+            for r in rows:
+                tags = r.tags or {}
+                source_detail = tags.get("source_detail") if isinstance(tags, dict) else None
+                results.append({
+                    "id": r.id,
+                    "title": r.title,
+                    "source_type": r.source_type,
+                    "author": r.author,
+                    "source_detail": source_detail,
+                    "snippet": r.snippet,
+                    "similarity": round(r.similarity, 4),
+                })
+            return results
+
+    def get_item_count(self) -> int:
+        with Session(engine) as session:
+            return len(session.exec(select(KnowledgeItem)).all())
+
+    def get_embedding_count(self) -> int:
+        with Session(engine) as session:
+            return len(session.exec(select(KnowledgeEmbedding)).all())
+
+    def get_grouped_items(self) -> list[dict]:
+        """
+        按 source_type → 二级字段（author / source_detail）两层聚合，返回树形结构数据。
+        按 SOURCE_TYPE_ORDER 固定顺序排列，仅显示有数据的大类（count > 0）。
+
+        二级节点 key 格式：`{source_type}|{二级字段名}|{字段值}`
+        例：manuscript|author|左鑫
+            military_report|source_detail|航空知识 2026年第1期
+
+        二级字段值为 null / 空字串 → 归入对应大类，不造空节点。
+        空大类（0条）不渲染。
+        """
+        with Session(engine) as session:
+            items = session.exec(select(KnowledgeItem)).all()
+
+        total_count = len(items)
+
+        # 按 source_type 分组：仅收集有数据的类别，再按固定顺序排列
+        type_groups: dict = {}
+        for item in items:
+            st = item.source_type or "manual"
+            if st not in type_groups:
+                type_groups[st] = []
+            type_groups[st].append(item)
+
+        # 只遍历有数据的类别，按 SOURCE_TYPE_ORDER 顺序
+        children = []
+        for st in self.SOURCE_TYPE_ORDER:
+            if st not in type_groups:
+                continue
+            st_items = type_groups[st]
+
+            secondary_field = self.SECONDARY_GROUP_FIELD.get(st)
+            grandchildren = []
+
+            if secondary_field is not None:
+                # 按二级字段分组
+                detail_groups: dict = {}
+                for item in st_items:
+                    if secondary_field == "source_detail":
+                        tags = item.tags or {}
+                        sd = tags.get("source_detail") if isinstance(tags, dict) else None
+                        field_val = sd
+                    else:
+                        field_val = (getattr(item, secondary_field, None) or "").strip() or None
+
+                    if field_val not in detail_groups:
+                        detail_groups[field_val] = []
+                    detail_groups[field_val].append(item)
+
+                for sd, sd_items in detail_groups.items():
+                    if sd is not None:
+                        grandchildren.append({
+                            "key": f"{st}|{secondary_field}|{sd}",
+                            "label": f"{sd}（{len(sd_items)}条）",
+                            "count": len(sd_items),
+                        })
+
+            children.append({
+                "key": st,
+                "label": f"{self.SOURCE_TYPE_LABEL.get(st, st)}（{len(st_items)}条）",
+                "count": len(st_items),
+                "children": grandchildren,
+            })
+
+        return [{
+            "key": "all",
+            "label": f"全部（{total_count}条）",
+            "count": total_count,
+            "children": children,
+        }]

backend/requirements.txt

@@ -0,0 +1,14 @@
+fastapi==0.111.0
+uvicorn[standard]==0.29.0
+sqlmodel==0.0.19
+psycopg2-binary==2.9.9
+pgvector==0.2.5
+passlib==1.7.4
+bcrypt==4.0.1
+itsdangerous==2.2.0
+python-multipart==0.0.9
+python-dotenv==1.0.1
+httpx==0.27.0
+python-frontmatter==1.1.0
+pandas>=2.0.0
+openpyxl>=3.1.0

backend/scripts/cleanup_test_data.py

@@ -0,0 +1,28 @@
+"""
+清理测试数据脚本
+"""
+
+from dotenv import load_dotenv
+from pathlib import Path
+
+_env_path = Path(__file__).parent.parent / ".env"
+load_dotenv(str(_env_path))
+
+from sqlalchemy import text
+from app.db.session import engine
+
+conn = engine.connect()
+
+# 清空知识库
+conn.execute(text("DELETE FROM knowledge_embeddings"))
+conn.execute(text("DELETE FROM knowledge_items"))
+conn.commit()
+
+# 查行数
+ki = conn.execute(text("SELECT COUNT(*) FROM knowledge_items")).scalar()
+ke = conn.execute(text("SELECT COUNT(*) FROM knowledge_embeddings")).scalar()
+
+print(f"knowledge_items rows: {ki}")
+print(f"knowledge_embeddings rows: {ke}")
+
+conn.close()

backend/scripts/create_admin.py

@@ -0,0 +1,81 @@
+"""
+创建管理员账号 — 交互式命令行脚本
+
+用法:
+    cd backend
+    py -3.12 scripts/create_admin.py
+"""
+
+import sys
+from pathlib import Path
+
+# 确保 backend/ 目录在 sys.path 中，使 app.* 导入正常工作
+backend_dir = Path(__file__).resolve().parent.parent
+sys.path.insert(0, str(backend_dir))
+
+from getpass import getpass
+
+from sqlmodel import Session, select
+
+from app.core.security import hash_password
+from app.db.session import engine
+from app.models.user import User, UserRole
+
+
+def main():
+    print("=" * 50)
+    print("  军事科技工作台 — 创建管理员账号")
+    print("=" * 50)
+    print()
+
+    username = input("请输入用户名: ").strip()
+    if not username:
+        print("❌ 用户名不能为空")
+        sys.exit(1)
+
+    display_name = input("请输入显示名（中文姓名）: ").strip()
+    if not display_name:
+        print("❌ 显示名不能为空")
+        sys.exit(1)
+
+    password = getpass("请输入密码（输入不显示）: ").strip()
+    if not password:
+        print("❌ 密码不能为空")
+        sys.exit(1)
+
+    password_confirm = getpass("请再次输入密码确认: ").strip()
+    if password != password_confirm:
+        print("❌ 两次输入的密码不一致")
+        sys.exit(1)
+
+    with Session(engine) as session:
+        # 检查用户名是否已存在
+        existing = session.exec(
+            select(User).where(User.username == username)
+        ).first()
+        if existing:
+            print(f"❌ 用户名「{username}」已存在，请换一个用户名")
+            sys.exit(1)
+
+        user = User(
+            username=username,
+            display_name=display_name,
+            password_hash=hash_password(password),
+            role=UserRole.zhipianren,
+            is_active=True,
+        )
+        session.add(user)
+        session.commit()
+        session.refresh(user)
+
+        print()
+        print("✅ 管理员账号创建成功！")
+        print(f"   用户名: {user.username}")
+        print(f"   显示名: {user.display_name}")
+        print(f"   角色:   {user.role.value}（制片人/管理员）")
+        print(f"   ID:     {user.id}")
+        print()
+
+
+if __name__ == "__main__":
+    main()

backend/scripts/import_smoke.py

@@ -0,0 +1,265 @@
+"""
+Task 2 批量导入冒烟脚本 — httpx（幂等版，可重复跑）
+确保后端已启动: py -3.12 -c "import sys; sys.path.insert(0, r'E:\tps-dashboard\backend'); import uvicorn; uvicorn.run('app.main:app', host='127.0.0.1', port=8000, reload=False)"
+
+Phase 2 Task 2 实测清单覆盖:
+  Step 1: GET  /api/imports/template       → 模板下载
+  Step 2: POST /api/imports/episodes       → 合法行单行导入
+  Step 3: POST /api/imports/episodes       → 部分失败行（成功+失败混合格式）
+  Step 4: GET  /api/imports/errors/{id}    → 失败行 Excel 下载
+  Step 5: POST /api/imports/episodes       → 2024年份缺失拦截 → 400
+  Step 6: POST /api/imports/episodes       → 重复期次拦截 → 409
+  Step 7: POST /api/imports/episodes       → 未知编导软落地 → 200 + editor_id=null
+"""
+
+import io
+import httpx
+import random
+from openpyxl import Workbook
+
+BASE = "http://localhost:8000"
+
+
+def login(username: str = "simonkoson", password: str = "liutong65") -> str:
+    r = httpx.post(f"{BASE}/api/auth/login", json={"username": username, "password": password})
+    r.raise_for_status()
+    return r.cookies.get("milsci_session", "")
+
+
+def step(name: str, expected_status: int, method: str, url: str, **kwargs):
+    cookies = kwargs.pop("cookies", {})
+    if method == "GET":
+        r = httpx.get(url, cookies=cookies, timeout=30.0, **kwargs)
+    elif method == "POST":
+        r = httpx.post(url, cookies=cookies, timeout=30.0, **kwargs)
+    elif method == "DELETE":
+        r = httpx.delete(url, cookies=cookies, timeout=30.0, **kwargs)
+    else:
+        raise ValueError(method)
+    status_ok = r.status_code == expected_status
+    mark = "[OK]" if status_ok else f"[FAIL expected {expected_status} got {r.status_code}]"
+    print(f"  {mark} [{method}] {name}")
+    if not status_ok:
+        print(f"     Response: {r.text[:300]}")
+    return r, status_ok
+
+
+def make_excel(rows: list[list]) -> bytes:
+    """构建简单 xlsx，rows[0] 为表头，后续为数据。"""
+    wb = Workbook()
+    ws = wb.active
+    for row in rows:
+        ws.append(row)
+    output = io.BytesIO()
+    wb.save(output)
+    output.seek(0)
+    return output.read()
+
+
+def cleanup_test_episodes(cookies: dict):
+    """清理测试用 episode_number（幂等清理）。"""
+    r = httpx.get(f"{BASE}/api/episodes?limit=100", cookies=cookies)
+    if r.status_code == 200:
+        for ep in r.json():
+            num = ep.get("episode_number", 0)
+            # 清理本次测试可能用到的期号段
+            if num in (99, 100, 101, 102, 200, 301):
+                httpx.delete(f"{BASE}/api/episodes/{ep['id']}", cookies=cookies)
+
+
+def main():
+    print("=== Login ===")
+    session_cookie = login()
+    if not session_cookie:
+        print("WARNING: session cookie empty")
+        session_cookie = ""
+    cookies = {"milsci_session": session_cookie}
+
+    # 先清理旧测试数据（幂等）
+    print("\n=== Cleanup old test episodes ===")
+    cleanup_test_episodes(cookies)
+
+    suffix = random.randint(1000, 9999)
+
+    # ---- Step 1: 模板下载 ----
+    print("\n=== Step 1: Template Download ===")
+    r, ok = step(
+        "GET /api/imports/template?type=episodes",
+        200,
+        "GET",
+        f"{BASE}/api/imports/template?type=episodes",
+        cookies=cookies,
+    )
+    content_type = r.headers.get("content-type", "")
+    is_binary = "application/vnd.openxmlformats" in content_type
+    print(f"  {'[OK]' if is_binary else '[FAIL]'} Content-Type: {content_type}")
+    print(f"  [INFO] Template downloaded ({len(r.content)} bytes)")
+
+    # ---- Step 2: 合法单行导入 ----
+    print("\n=== Step 2: Normal import (1 valid row) ===")
+    excel_data = make_excel([
+        ["episode_number", "program_name", "air_date", "editor_name",
+         "audience_share", "audience_rating", "is_rerun", "notes"],
+        [99, "《军事科技》实测专用", "2026-06-01", "张颖",
+         0.75, 3.8, "否", "Plan smoke test"],
+    ])
+    r, ok = step(
+        "POST /api/imports/episodes (1 valid row) → 200",
+        200,
+        "POST",
+        f"{BASE}/api/imports/episodes",
+        files={"file": ("test_valid.xlsx", excel_data, "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet")},
+        cookies=cookies,
+    )
+    if ok:
+        result = r.json()
+        print(f"  [INFO] total={result['total_rows']} success={result['success_count']} failed={result['failed_count']}")
+        ok = (result["success_count"] == 1 and result["failed_count"] == 0)
+        print(f"  {'[OK]' if ok else '[FAIL]'} Row count correct")
+
+    # ---- Step 3: 部分失败行回显 ----
+    print("\n=== Step 3: Partial failure (3 rows: 1 ok + 2 bad) ===")
+    excel_data = make_excel([
+        ["episode_number", "program_name", "air_date", "editor_name",
+         "audience_share", "audience_rating", "is_rerun", "notes"],
+        [100, "《军事科技》成功行", "2026-07-01", "张颖", 0.71, 3.5, "否", ""],
+        [101, "无效日期格式行", "不是日期", "张颖", 0.72, 3.6, "否", ""],
+        [102, "非整数期号行", "invalid-ep-num", "张颖", 0.73, 3.7, "否", ""],
+    ])
+    r, ok = step(
+        "POST /api/imports/episodes (1ok+2bad) → 200 with errors",
+        200,
+        "POST",
+        f"{BASE}/api/imports/episodes",
+        files={"file": ("test_mixed.xlsx", excel_data, "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet")},
+        cookies=cookies,
+    )
+    batch_id = None
+    if ok:
+        result = r.json()
+        print(f"  [INFO] total={result['total_rows']} success={result['success_count']} failed={result['failed_count']}")
+        ok_count = (result["success_count"] == 1 and result["failed_count"] == 2)
+        print(f"  {'[OK]' if ok_count else '[FAIL]'} Failure count correct")
+        # 检查 row_number 是否对齐 Excel 行号（第2行=Excel第2行，第3行=Excel第3行，第4行=Excel第4行）
+        if result["errors"]:
+            row_nums = sorted([e["row_number"] for e in result["errors"]])
+            print(f"  [INFO] error row_numbers: {row_nums}")
+            ok_rows = (row_nums == [3, 4])
+            print(f"  {'[OK]' if ok_rows else '[FAIL]'} row_number对齐Excel自然行号(表头第1行,数据从第2行起算)")
+        batch_id = result.get("batch_id")
+        if batch_id:
+            print(f"  [INFO] batch_id={batch_id}")
+
+    # ---- Step 4: 失败行下载 ----
+    print("\n=== Step 4: Error Excel Download ===")
+    if batch_id:
+        r, ok = step(
+            f"GET /api/imports/errors/{batch_id} → 200 binary",
+            200,
+            "GET",
+            f"{BASE}/api/imports/errors/{batch_id}",
+            cookies=cookies,
+        )
+        ct = r.headers.get("content-type", "")
+        is_xlsx = "application/vnd.openxmlformats" in ct
+        print(f"  {'[OK]' if is_xlsx else '[FAIL]'} Content-Type: {ct}")
+        print(f"  [INFO] Error Excel downloaded ({len(r.content)} bytes)")
+    else:
+        print("  [SKIP] No batch_id from step 3")
+
+    # ---- Step 5: 年份目标缺失拦截 ----
+    print("\n=== Step 5: Missing yearly target → 400 ===")
+    excel_data = make_excel([
+        ["episode_number", "program_name", "air_date", "editor_name",
+         "audience_share", "audience_rating", "is_rerun", "notes"],
+        [200, "《军事科技》跨年测试", "2024-01-15", "张颖", 0.68, 3.2, "否", ""],
+    ])
+    r, ok = step(
+        "POST with 2024 air_date (no 2024 target) → 400",
+        400,
+        "POST",
+        f"{BASE}/api/imports/episodes",
+        files={"file": ("test_2024.xlsx", excel_data, "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet")},
+        cookies=cookies,
+    )
+    if ok:
+        detail = r.json().get("detail", "")
+        has_year_target = "年度目标" in detail and "2024" in detail
+        print(f"  {'[OK]' if has_year_target else '[FAIL]'} detail包含'年度目标'+'2024': {detail[:100]}")
+
+    # ---- Step 6: 重复期次拦截 ----
+    print("\n=== Step 6: Duplicate episode_number ===")
+    excel_data = make_excel([
+        ["episode_number", "program_name", "air_date", "editor_name",
+         "audience_share", "audience_rating", "is_rerun", "notes"],
+        [99, "《军事科技》重复期次", "2026-06-15", "张颖", 0.76, 3.9, "否", ""],
+    ])
+    r, ok = step(
+        "POST episode_number=99 (duplicate in same year) → 409 or row-level rejection",
+        200,
+        "POST",
+        f"{BASE}/api/imports/episodes",
+        files={"file": ("test_duplicate.xlsx", excel_data, "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet")},
+        cookies=cookies,
+    )
+    if ok:
+        result = r.json()
+        # 数据库唯一约束触发：整行被 reject，success=0, failed=1
+        has_db_error = result["failed_count"] == 1 and result["success_count"] == 0
+        has_unique_violation = any("UniqueViolation" in e.get("reason", "") or "idx_episodes_year_number" in e.get("reason", "") for e in result["errors"])
+        print(f"  {'[OK]' if has_unique_violation else '[FAIL]'} 数据库唯一约束触发 (failed_count=1, UniqueViolation in reason)")
+
+    # ---- Step 7: 未知编导软落地 ----
+    print("\n=== Step 7: Unknown editor soft-landing ===")
+    excel_data = make_excel([
+        ["episode_number", "program_name", "air_date", "editor_name",
+         "audience_share", "audience_rating", "is_rerun", "notes"],
+        [301, "《军事科技》编导软引用测试", "2026-09-01", "李不存在", 0.69, 3.1, "否", ""],
+    ])
+    r, ok = step(
+        "POST with unknown editor_name='李不存在' → 200",
+        200,
+        "POST",
+        f"{BASE}/api/imports/episodes",
+        files={"file": ("test_unknown_editor.xlsx", excel_data, "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet")},
+        cookies=cookies,
+    )
+    if ok:
+        result = r.json()
+        ok = (result["success_count"] == 1 and result["failed_count"] == 0)
+        print(f"  {'[OK]' if ok else '[FAIL]'} success_count=1")
+    # 验证 editor_id=null：先从列表拿到真实 id，再调用 GET /api/episodes/{id}
+    r_list, ok_list = step(
+        "GET /api/episodes?limit=100 → 找到刚导入的 episode_number=301",
+        200,
+        "GET",
+        f"{BASE}/api/episodes?limit=100",
+        cookies=cookies,
+    )
+    if ok_list:
+        eps = r_list.json()
+        ep_301 = next((e for e in eps if e.get("episode_number") == 301), None)
+        if ep_301:
+            real_id = ep_301["id"]
+            r_ep, ok_ep = step(
+                f"GET /api/episodes/{real_id} → editor_id=null, editor_name_snapshot='李不存在'",
+                200,
+                "GET",
+                f"{BASE}/api/episodes/{real_id}",
+                cookies=cookies,
+            )
+            if ok_ep:
+                ep = r_ep.json()
+                editor_id_null = ep.get("editor_id") is None
+                name_match = ep.get("editor_name_snapshot") == "李不存在"
+                print(f"  {'[OK]' if editor_id_null else '[FAIL]'} editor_id=null ({ep.get('editor_id')})")
+                print(f"  {'[OK]' if name_match else '[FAIL]'} editor_name_snapshot='李不存在' ({ep.get('editor_name_snapshot')})")
+        else:
+            print(f"  [FAIL] episode_number=301 not found in list (import may have failed)")
+
+    print("\n=== All steps done ===")
+    print("NOTE: This script is辅助. 制片人仍需在 Swagger UI (http://localhost:8000/docs) 手动跑完 8 步确认全绿.")
+
+
+if __name__ == "__main__":
+    main()

backend/scripts/smoke_test.py

@@ -0,0 +1,157 @@
+"""
+Task 1 冒烟测试脚本 — httpx（幂等版，可重复跑）
+确保后端已启动: uvicorn app.main:app --reload
+"""
+
+import httpx
+import random
+
+BASE = "http://localhost:8000"
+HEADERS = {"Content-Type": "application/json"}
+
+
+def login(username: str, password: str) -> str:
+    r = httpx.post(f"{BASE}/api/auth/login", json={"username": username, "password": password})
+    r.raise_for_status()
+    return r.cookies.get("milsci_session", "")
+
+
+def step(name: str, expected_status: int, method: str, url: str, **kwargs):
+    cookies = kwargs.pop("cookies", {})
+    if method == "GET":
+        r = httpx.get(url, cookies=cookies, **kwargs)
+    elif method == "POST":
+        r = httpx.post(url, cookies=cookies, **kwargs)
+    elif method == "PATCH":
+        r = httpx.patch(url, cookies=cookies, **kwargs)
+    elif method == "DELETE":
+        r = httpx.delete(url, cookies=cookies, **kwargs)
+    else:
+        raise ValueError(method)
+    status_ok = r.status_code == expected_status
+    mark = "[OK]" if status_ok else f"[FAIL expected {expected_status} got {r.status_code}]"
+    print(f"{mark} [{method}] {name}")
+    if not status_ok:
+        print(f"   Response: {r.text[:300]}")
+    return r, status_ok
+
+
+def main():
+    print("=== Login ===")
+    session_cookie = login("simonkoson", "liutong65")
+    if not session_cookie:
+        print("WARNING: session cookie empty, trying anyway")
+        session_cookie = ""
+    cookies = {"milsci_session": session_cookie}
+
+    # 使用随机后缀避免与上次数据冲突
+    suffix = random.randint(1000, 9999)
+
+    print("\n=== Step 1: Create zebian account ===")
+    r, ok = step(
+        "POST /api/users create zebian",
+        201,
+        "POST",
+        f"{BASE}/api/users",
+        json={
+            "username": f"test_zebian_{suffix}",
+            "display_name": f"Test Zebra {suffix}",
+            "password": "Test1234",
+            "role": "zebian",
+        },
+        cookies=cookies,
+    )
+    zebian_id = r.json().get("id") if ok else None
+    if ok:
+        print(f"   zebian created, ID={zebian_id}")
+
+    print("\n=== Step 2: Create 2026 yearly target (idempotent) ===")
+    # 先尝试删已存在的（方便重跑）
+    r_list = httpx.get(f"{BASE}/api/yearly_targets", cookies=cookies)
+    if r_list.status_code == 200:
+        for t in r_list.json():
+            if t.get("year") == 2026:
+                del_id = t["id"]
+                httpx.delete(f"{BASE}/api/yearly_targets/{del_id}", cookies=cookies)
+                print(f"   cleaned up existing 2026 target ID={del_id}")
+    # 再建新的
+    r, ok = step(
+        "POST /api/yearly_targets create 2026",
+        201,
+        "POST",
+        f"{BASE}/api/yearly_targets",
+        json={"year": 2026, "base_target": 0.65, "stretch_target": 0.90},
+        cookies=cookies,
+    )
+    target_2026_id = r.json().get("id") if ok else None
+    if ok:
+        print(f"   target created, ID={target_2026_id}")
+
+    print("\n=== Step 3: Create 2026 again (should be 409) ===")
+    step(
+        "POST /api/yearly_targets duplicate 2026",
+        409,
+        "POST",
+        f"{BASE}/api/yearly_targets",
+        json={"year": 2026, "base_target": 0.70, "stretch_target": 0.95},
+        cookies=cookies,
+    )
+
+    print("\n=== Step 4: Create episode with editor_id=null (idempotent) ===")
+    # 先清理 episode_number
+    r_list = httpx.get(f"{BASE}/api/episodes?limit=100", cookies=cookies)
+    if r_list.status_code == 200:
+        for ep in r_list.json():
+            if ep.get("episode_number") == 9999:
+                httpx.delete(f"{BASE}/api/episodes/{ep['id']}", cookies=cookies)
+                print(f"   cleaned up existing episode 9999 ID={ep['id']}")
+    r, ok = step(
+        "POST /api/episodes editor_id=null",
+        201,
+        "POST",
+        f"{BASE}/api/episodes",
+        json={
+            "episode_number": 9999,
+            "program_name": "Test Episode Smoke",
+            "air_date": "2026-05-15",
+            "editor_id": None,
+            "editor_name_snapshot": "Test Director Zhang",
+            "audience_share": 0.72,
+        },
+        cookies=cookies,
+    )
+    episode_9999_id = r.json().get("id") if ok else None
+    if ok:
+        print(f"   episode created, ID={episode_9999_id}")
+
+    print("\n=== Step 5: GET /api/episodes list ===")
+    step(
+        "GET /api/episodes list",
+        200,
+        "GET",
+        f"{BASE}/api/episodes?limit=10",
+        cookies=cookies,
+    )
+
+    print("\n=== Step 6: PATCH yearly_target try to change year ===")
+    if target_2026_id:
+        r, ok = step(
+            "PATCH try to change year field",
+            200,
+            "PATCH",
+            f"{BASE}/api/yearly_targets/{target_2026_id}",
+            json={"year": 2027, "base_target": 0.65, "stretch_target": 0.90},
+            cookies=cookies,
+        )
+        if ok:
+            resp_year = r.json().get("year")
+            if resp_year == 2026:
+                print(f"   [OK] year field unchanged, still 2026 (correct: year blocked)")
+            else:
+                print(f"   [FAIL] year changed to {resp_year} (should be blocked)")
+
+    print("\n=== All steps done ===")
+
+
+if __name__ == "__main__":
+    main()

backend/scripts/smoke_yearly_targets_auth.py

@@ -0,0 +1,419 @@
+"""
+yearly_targets 写权限收紧 — 冒烟测试脚本
+
+幂等，可重复跑。
+确保后端已启动: uvicorn app.main:app --reload
+
+12 步覆盖:
+1. zhipianren POST → 201
+2. zebian POST → 201
+3. biandao POST → 403
+4. 未登录 POST → 401
+5. zhipianren PATCH → 200
+6. zebian PATCH → 200
+7. biandao PATCH → 403
+8. zhipianren DELETE → 200
+9. zebian DELETE → 403
+10. biandao DELETE → 403
+11. biandao GET 列表 → 200
+12. zhipianren GET 列表 → 200
+"""
+
+import sys
+from pathlib import Path
+
+backend_dir = Path(__file__).resolve().parent.parent
+sys.path.insert(0, str(backend_dir))
+
+import httpx
+
+BASE = "http://localhost:8000"
+HEADERS = {"Content-Type": "application/json"}
+
+
+def login(username: str, password: str) -> str:
+    """登录并返回 session cookie。"""
+    r = httpx.post(f"{BASE}/api/auth/login", json={"username": username, "password": password})
+    r.raise_for_status()
+    return r.cookies.get("milsci_session", "")
+
+
+def step(name: str, expected_status: int, method: str, url: str, **kwargs):
+    """执行一个 HTTP 步骤，比较实际状态码与期望值。"""
+    cookies = kwargs.pop("cookies", {})
+    if method == "GET":
+        r = httpx.get(url, cookies=cookies, timeout=10)
+    elif method == "POST":
+        r = httpx.post(url, cookies=cookies, headers=HEADERS, timeout=10, **kwargs)
+    elif method == "PATCH":
+        r = httpx.patch(url, cookies=cookies, headers=HEADERS, timeout=10, **kwargs)
+    elif method == "DELETE":
+        r = httpx.delete(url, cookies=cookies, timeout=10)
+    else:
+        raise ValueError(method)
+
+    status_ok = r.status_code == expected_status
+    mark = "[OK]" if status_ok else f"[FAIL expected {expected_status} got {r.status_code}]"
+    print(f"  {mark} [{method}] {name}")
+    if not status_ok:
+        print(f"       Response: {r.text[:300]}")
+    return r, status_ok
+
+
+def ensure_user(cookies: dict, username: str, display_name: str, role: str, password: str = "Test1234"):
+    """幂等创建测试账号（如果不存在）。"""
+    # 先尝试登录（账号可能已存在）
+    try:
+        r = httpx.get(f"{BASE}/api/users/me", cookies=cookies, timeout=5)
+        if r.status_code == 200:
+            me = r.json()
+            if me.get("username") == username:
+                print(f"  [SKIP] user {username} already exists, reusing")
+                return
+    except Exception:
+        pass
+
+    # 尝试创建
+    r = httpx.post(
+        f"{BASE}/api/users",
+        json={"username": username, "display_name": display_name, "password": password, "role": role},
+        cookies=cookies,
+        headers=HEADERS,
+        timeout=10,
+    )
+    if r.status_code == 201:
+        print(f"  [OK] created {username} ({role})")
+    elif r.status_code == 409:
+        print(f"  [SKIP] {username} already exists")
+    else:
+        print(f"  [WARN] create {username} returned {r.status_code}: {r.text[:100]}")
+
+
+def cleanup_test_target(cookies: dict):
+    """清理测试用年度目标（2027），幂等。"""
+    try:
+        r = httpx.get(f"{BASE}/api/yearly_targets", cookies=cookies, timeout=10)
+        if r.status_code == 200:
+            for t in r.json():
+                if t.get("year") == 2027:
+                    httpx.delete(f"{BASE}/api/yearly_targets/{t['id']}", cookies=cookies, timeout=10)
+                    print(f"  [CLEAN] removed existing 2027 target ID={t['id']}")
+    except Exception:
+        pass
+
+
+def main():
+    print("=" * 60)
+    print("  yearly_targets 写权限收紧 — 冒烟测试")
+    print("=" * 60)
+
+    # Step 0: 用 zhipianren 登录，准备测试账号
+    print("\n[Step 0] Login as zhipianren and ensure test accounts exist")
+    try:
+        admin_cookie = login("simonkoson", "liutong65")
+    except Exception as e:
+        print(f"  [FAIL] cannot login as simonkoson: {e}")
+        return
+
+    admin_cookies = {"milsci_session": admin_cookie}
+    print("  [OK] logged in as simonkoson (zhipianren)")
+
+    # 幂等创建 zebian / biandao 测试账号
+    ensure_user(admin_cookies, "test_zebian_smoke", "TestZebra Smoke", "zebian")
+    ensure_user(admin_cookies, "test_biandao_smoke", "TestBiandao Smoke", "biandao")
+
+    # 登录 zebian 和 biandao 获取 cookie
+    print("\n[Step 0b] Login as zebian and biandao")
+    zebian_cookie = login("test_zebian_smoke", "Test1234")
+    zebian_cookies = {"milsci_session": zebian_cookie}
+    print("  [OK] logged in as test_zebian_smoke (zebian)")
+
+    biandao_cookie = login("test_biandao_smoke", "Test1234")
+    biandao_cookies = {"milsci_session": biandao_cookie}
+    print("  [OK] logged in as test_biandao_smoke (biandao)")
+
+    # 清理可能残留的 2027 目标
+    cleanup_test_target(admin_cookies)
+
+    results = []
+
+    # =====================================================================
+    # Step 1: zhipianren POST → 201
+    # =====================================================================
+    print("\n[Step 1] POST new yearly target — zhipianren → 201")
+    r, ok = step(
+        "POST /api/yearly_targets (zhipianren)",
+        201,
+        "POST",
+        f"{BASE}/api/yearly_targets",
+        json={"year": 2027, "base_target": 0.66, "stretch_target": 0.91},
+        cookies=admin_cookies,
+    )
+    target_id = r.json().get("id") if ok else None
+    results.append(("Step 1: zhipianren POST 201", ok))
+
+    # =====================================================================
+    # Step 2: zebian POST → 201
+    # =====================================================================
+    print("\n[Step 2] POST new yearly target — zebian → 201")
+    # 清理 2027（刚被 zhipianren 创建了）
+    cleanup_test_target(admin_cookies)
+    r, ok = step(
+        "POST /api/yearly_targets (zebian)",
+        201,
+        "POST",
+        f"{BASE}/api/yearly_targets",
+        json={"year": 2027, "base_target": 0.66, "stretch_target": 0.91},
+        cookies=zebian_cookies,
+    )
+    target_id2 = r.json().get("id") if ok else None
+    results.append(("Step 2: zebian POST 201", ok))
+
+    # =====================================================================
+    # Step 3: biandao POST → 403
+    # =====================================================================
+    print("\n[Step 3] POST new yearly target — biandao → 403")
+    cleanup_test_target(admin_cookies)
+    r, ok = step(
+        "POST /api/yearly_targets (biandao)",
+        403,
+        "POST",
+        f"{BASE}/api/yearly_targets",
+        json={"year": 2027, "base_target": 0.66, "stretch_target": 0.91},
+        cookies=biandao_cookies,
+    )
+    results.append(("Step 3: biandao POST 403", ok))
+
+    # =====================================================================
+    # Step 4: 未登录 POST → 401
+    # =====================================================================
+    print("\n[Step 4] POST new yearly target — not logged in → 401")
+    r, ok = step(
+        "POST /api/yearly_targets (no auth)",
+        401,
+        "POST",
+        f"{BASE}/api/yearly_targets",
+        json={"year": 2027, "base_target": 0.66, "stretch_target": 0.91},
+        cookies={},
+    )
+    results.append(("Step 4: no auth POST 401", ok))
+
+    # =====================================================================
+    # Step 5: zhipianren PATCH → 200
+    # =====================================================================
+    print("\n[Step 5] PATCH yearly target — zhipianren → 200")
+    # 先用 admin 建一个可 PATCH 的目标
+    cleanup_test_target(admin_cookies)
+    r = httpx.post(
+        f"{BASE}/api/yearly_targets",
+        json={"year": 2027, "base_target": 0.66, "stretch_target": 0.91},
+        cookies=admin_cookies,
+        headers=HEADERS,
+        timeout=10,
+    )
+    patch_id = r.json().get("id") if r.status_code == 201 else None
+
+    if patch_id:
+        r, ok = step(
+            "PATCH /api/yearly_targets/{id} (zhipianren)",
+            200,
+            "PATCH",
+            f"{BASE}/api/yearly_targets/{patch_id}",
+            json={"base_target": 0.67},
+            cookies=admin_cookies,
+        )
+    else:
+        ok = False
+        print("  [FAIL] could not create target for PATCH test")
+    results.append(("Step 5: zhipianren PATCH 200", ok))
+
+    # =====================================================================
+    # Step 6: zebian PATCH → 200
+    # =====================================================================
+    print("\n[Step 6] PATCH yearly target — zebian → 200")
+    cleanup_test_target(admin_cookies)
+    r = httpx.post(
+        f"{BASE}/api/yearly_targets",
+        json={"year": 2027, "base_target": 0.66, "stretch_target": 0.91},
+        cookies=admin_cookies,
+        headers=HEADERS,
+        timeout=10,
+    )
+    patch_id2 = r.json().get("id") if r.status_code == 201 else None
+
+    if patch_id2:
+        r, ok = step(
+            "PATCH /api/yearly_targets/{id} (zebian)",
+            200,
+            "PATCH",
+            f"{BASE}/api/yearly_targets/{patch_id2}",
+            json={"stretch_target": 0.92},
+            cookies=zebian_cookies,
+        )
+    else:
+        ok = False
+        print("  [FAIL] could not create target for PATCH test")
+    results.append(("Step 6: zebian PATCH 200", ok))
+
+    # =====================================================================
+    # Step 7: biandao PATCH → 403
+    # =====================================================================
+    print("\n[Step 7] PATCH yearly target — biandao → 403")
+    cleanup_test_target(admin_cookies)
+    r = httpx.post(
+        f"{BASE}/api/yearly_targets",
+        json={"year": 2027, "base_target": 0.66, "stretch_target": 0.91},
+        cookies=admin_cookies,
+        headers=HEADERS,
+        timeout=10,
+    )
+    patch_id3 = r.json().get("id") if r.status_code == 201 else None
+
+    if patch_id3:
+        r, ok = step(
+            "PATCH /api/yearly_targets/{id} (biandao)",
+            403,
+            "PATCH",
+            f"{BASE}/api/yearly_targets/{patch_id3}",
+            json={"base_target": 0.68},
+            cookies=biandao_cookies,
+        )
+    else:
+        ok = False
+        print("  [FAIL] could not create target for PATCH test")
+    results.append(("Step 7: biandao PATCH 403", ok))
+
+    # =====================================================================
+    # Step 8: zhipianren DELETE → 200
+    # =====================================================================
+    print("\n[Step 8] DELETE yearly target — zhipianren → 200")
+    cleanup_test_target(admin_cookies)
+    r = httpx.post(
+        f"{BASE}/api/yearly_targets",
+        json={"year": 2027, "base_target": 0.66, "stretch_target": 0.91},
+        cookies=admin_cookies,
+        headers=HEADERS,
+        timeout=10,
+    )
+    del_id = r.json().get("id") if r.status_code == 201 else None
+
+    if del_id:
+        r, ok = step(
+            "DELETE /api/yearly_targets/{id} (zhipianren)",
+            204,
+            "DELETE",
+            f"{BASE}/api/yearly_targets/{del_id}",
+            cookies=admin_cookies,
+        )
+    else:
+        ok = False
+        print("  [FAIL] could not create target for DELETE test")
+    results.append(("Step 8: zhipianren DELETE 204", ok))
+
+    # =====================================================================
+    # Step 9: zebian DELETE → 403
+    # =====================================================================
+    print("\n[Step 9] DELETE yearly target — zebian → 403")
+    cleanup_test_target(admin_cookies)
+    r = httpx.post(
+        f"{BASE}/api/yearly_targets",
+        json={"year": 2027, "base_target": 0.66, "stretch_target": 0.91},
+        cookies=admin_cookies,
+        headers=HEADERS,
+        timeout=10,
+    )
+    del_id2 = r.json().get("id") if r.status_code == 201 else None
+
+    if del_id2:
+        r, ok = step(
+            "DELETE /api/yearly_targets/{id} (zebian)",
+            403,
+            "DELETE",
+            f"{BASE}/api/yearly_targets/{del_id2}",
+            cookies=zebian_cookies,
+        )
+    else:
+        ok = False
+        print("  [FAIL] could not create target for DELETE test")
+    results.append(("Step 9: zebian DELETE 403", ok))
+
+    # =====================================================================
+    # Step 10: biandao DELETE → 403
+    # =====================================================================
+    print("\n[Step 10] DELETE yearly target — biandao → 403")
+    cleanup_test_target(admin_cookies)
+    r = httpx.post(
+        f"{BASE}/api/yearly_targets",
+        json={"year": 2027, "base_target": 0.66, "stretch_target": 0.91},
+        cookies=admin_cookies,
+        headers=HEADERS,
+        timeout=10,
+    )
+    del_id3 = r.json().get("id") if r.status_code == 201 else None
+
+    if del_id3:
+        r, ok = step(
+            "DELETE /api/yearly_targets/{id} (biandao)",
+            403,
+            "DELETE",
+            f"{BASE}/api/yearly_targets/{del_id3}",
+            cookies=biandao_cookies,
+        )
+    else:
+        ok = False
+        print("  [FAIL] could not create target for DELETE test")
+    results.append(("Step 10: biandao DELETE 403", ok))
+
+    # =====================================================================
+    # Step 11: biandao GET 列表 → 200
+    # =====================================================================
+    print("\n[Step 11] GET yearly targets list — biandao → 200")
+    r, ok = step(
+        "GET /api/yearly_targets (biandao)",
+        200,
+        "GET",
+        f"{BASE}/api/yearly_targets",
+        cookies=biandao_cookies,
+    )
+    results.append(("Step 11: biandao GET 200", ok))
+
+    # =====================================================================
+    # Step 12: zhipianren GET 列表 → 200
+    # =====================================================================
+    print("\n[Step 12] GET yearly targets list — zhipianren → 200")
+    r, ok = step(
+        "GET /api/yearly_targets (zhipianren)",
+        200,
+        "GET",
+        f"{BASE}/api/yearly_targets",
+        cookies=admin_cookies,
+    )
+    results.append(("Step 12: zhipianren GET 200", ok))
+
+    # =====================================================================
+    # Summary
+    # =====================================================================
+    print("\n" + "=" * 60)
+    print("  结果汇总")
+    print("=" * 60)
+    passed = sum(1 for _, ok in results if ok)
+    total = len(results)
+    for name, ok in results:
+        mark = "[OK]" if ok else "[FAIL]"
+        print(f"  {mark} {name}")
+
+    print(f"\n  PASS: {passed}/{total}")
+    if passed == total:
+        print("  ALL PASS")
+    else:
+        print("  SOME FAILURES")
+
+    print("\n测试账号已保留(供后续手动清理):")
+    print("  - test_zebian_smoke (zebian)")
+    print("  - test_biandao_smoke (biandao)")
+
+    print("\n" + "=" * 60)
+
+
+if __name__ == "__main__":
+    main()

backend/scripts/test_embo01_api.py

@@ -0,0 +1,70 @@
+"""
+探路脚本 — 调 MiniMax embo-01，打印原始返回 JSON
+确认向量字段位置和维度后再写正式 service。
+"""
+
+import httpx
+import json
+import os
+from pathlib import Path
+
+# 加载 .env
+from dotenv import load_dotenv
+_env_path = Path(__file__).parent.parent / ".env"
+load_dotenv(str(_env_path))
+
+api_key = os.environ.get("MINIMAX_EMBED_API_KEY", "")
+group_id = os.environ.get("MINIMAX_GROUP_ID", "")
+
+if not api_key or api_key == "your_api_key_here":
+    print("[ERROR] MINIMAX_EMBED_API_KEY not configured, please edit backend/.env")
+    exit(1)
+if not group_id or group_id == "your_group_id_here":
+    print("[ERROR] MINIMAX_GROUP_ID not configured, please edit backend/.env")
+    exit(1)
+
+print(f"API Key (first 4 chars): {api_key[:4]}...")
+print(f"GroupId: {group_id}")
+print()
+
+# 最小调用
+test_text = "这是一段测试文本，用于验证 embo-01 接口返回结构。"
+
+print(f"Sending request, test text: {test_text}")
+print("-" * 60)
+
+try:
+    resp = httpx.post(
+        "https://api.minimax.chat/v1/embeddings",
+        headers={
+            "Authorization": f"Bearer {api_key}",
+            "GroupId": group_id,
+            "Content-Type": "application/json",
+        },
+        json={"model": "embo-01", "texts": [test_text], "type": "db"},
+        timeout=30.0,
+    )
+    print(f"HTTP status: {resp.status_code}")
+    print()
+    data = resp.json()
+    print(json.dumps(data, indent=2, ensure_ascii=False))
+
+    # 提取向量，验证维度
+    print()
+    print("-" * 60)
+    vectors = data.get("vectors", [])
+    if vectors and len(vectors) > 0:
+        embedding = vectors[0]
+        dim = len(embedding)
+        print(f"[OK] Embedding field: vectors[0]")
+        print(f"[OK] Embedding dimension: {dim}")
+        if dim != 1536:
+            print(f"[STOP] Dimension is NOT 1536! Got {dim} - stopping here")
+        else:
+            print(f"[OK] Dimension correct: 1536")
+            print(f"[OK] API call successful, structure confirmed.")
+    else:
+        print("[WARNING] vectors not found in response")
+
+except Exception as e:
+    print(f"[ERROR] Request failed: {e}")

backend/scripts/verify_embedding.py

@@ -0,0 +1,79 @@
+"""
+全链路验证脚本 — TPS 知识库 embedding 最小链路
+
+验证步骤：
+1. 读取 backend/sample_md/ 下的 5 篇 .md 文件
+2. 调用 embo-01 转成向量（打印维度）
+3. 存入 knowledge_items + knowledge_embeddings（打印行数）
+4. 执行语义检索（打印查询句 + 最相似笔记）
+5. 查 episodes 表行数（打印，只读不动）
+"""
+
+import os
+from pathlib import Path
+
+from dotenv import load_dotenv
+from sqlmodel import text
+
+# 加载 .env
+_env_path = Path(__file__).parent.parent / ".env"
+load_dotenv(str(_env_path))
+
+from app.services.knowledge_service import KnowledgeService
+from app.db.session import engine
+
+
+def main():
+    print("=" * 60)
+    print("TPS Knowledge Base — Embedding Full链路验证")
+    print("=" * 60)
+
+    sample_dir = Path(__file__).parent.parent / "sample_md"
+    md_files = sorted(sample_dir.glob("*.md"))
+    print(f"\n[FIND] Found {len(md_files)} .md files in sample_md/")
+
+    ks = KnowledgeService()
+
+    # 1. 写入知识库
+    print("\n[STEP 1] Storing MD files into knowledge base...")
+    items_stored = []
+    for mf in md_files:
+        title = mf.stem  # 文件名（不含扩展名）作为标题
+        content = mf.read_text(encoding="utf-8")
+        item = ks.store_md_file(
+            title=title,
+            content_md=content,
+            source_file_name=mf.name,
+            source_type="manual",
+        )
+        items_stored.append(item)
+        print(f"  - Stored: {item.title} (id={item.id})")
+
+    ki_count = ks.get_item_count()
+    ke_count = ks.get_embedding_count()
+    print(f"\n[OK] knowledge_items rows: {ki_count}")
+    print(f"[OK] knowledge_embeddings rows: {ke_count}")
+
+    # 2. 语义检索
+    print("\n[STEP 2] Semantic search test...")
+    query = "五代战斗机的隐身技术有哪些关键要素？"
+    print(f"Query: {query}")
+    results = ks.search_similar(query, top_k=3)
+    print(f"\n[OK] Top 3 similar notes:")
+    for i, r in enumerate(results, 1):
+        print(f"  {i}. [{r['similarity']}] {r['title']}")
+
+    # 3. 查 episodes 表行数（只读不动）
+    print("\n[STEP 3] Episodes table (read-only)...")
+    with engine.connect() as conn:
+        result = conn.execute(text("SELECT COUNT(*) FROM episodes"))
+        episode_count = result.scalar()
+    print(f"[OK] episodes table row count: {episode_count}")
+
+    print("\n" + "=" * 60)
+    print("Verification complete.")
+    print("=" * 60)
+
+
+if __name__ == "__main__":
+    main()

backend/sql/001_init.sql

@@ -0,0 +1,238 @@
+-- ============================================================
+-- 001_init.sql — 《军事科技》栏目内部工作台 数据库初始化脚本
+-- 一次性建完所有核心表骨架（Phase 1-4 全部表）
+-- 执行方式: psql -U postgres -d milsci_dev -f 001_init.sql
+-- ============================================================
+
+-- 扩展: pgvector（向量检索）
+CREATE EXTENSION IF NOT EXISTS vector;
+
+-- ============================================================
+-- 1. users — 用户表（Phase 1 用）
+-- ============================================================
+CREATE TABLE IF NOT EXISTS users (
+    id              SERIAL PRIMARY KEY,
+    username        VARCHAR(50)  NOT NULL UNIQUE,
+    display_name    VARCHAR(100) NOT NULL,
+    password_hash   VARCHAR(128) NOT NULL,
+    -- 角色: zhipianren=制片人/管理员, zebian=责编, biandao=编导
+    role            VARCHAR(20)  NOT NULL DEFAULT 'biandao'
+                    CHECK (role IN ('zhipianren', 'zebian', 'biandao')),
+    profile_text    TEXT,
+    is_active       BOOLEAN      NOT NULL DEFAULT TRUE,
+    created_at      TIMESTAMPTZ  NOT NULL DEFAULT NOW(),
+    updated_at      TIMESTAMPTZ  NOT NULL DEFAULT NOW()
+);
+
+CREATE INDEX idx_users_role ON users (role);
+
+-- ============================================================
+-- 2. episodes — 节目期次（Phase 2 用，字段完整按 dev_plan.md 8.1 节）
+-- ============================================================
+CREATE TABLE IF NOT EXISTS episodes (
+    id                      SERIAL PRIMARY KEY,
+    episode_number          INTEGER      NOT NULL UNIQUE,
+    program_name            VARCHAR(200) NOT NULL,
+    air_date                DATE         NOT NULL,
+    -- 编导（软引用 + 快照模式）
+    editor_id               INTEGER      REFERENCES users(id) ON DELETE SET NULL,
+    editor_name_snapshot    VARCHAR(100) NOT NULL,
+    -- 收视数据
+    audience_share          NUMERIC(6,4),
+    audience_rating         NUMERIC(6,4),
+    -- 重播相关
+    is_rerun                BOOLEAN      NOT NULL DEFAULT FALSE,
+    original_episode_id     INTEGER      REFERENCES episodes(id) ON DELETE SET NULL,
+    is_time_sensitive       BOOLEAN      NOT NULL DEFAULT FALSE,
+    -- 外拍信息
+    is_outdoor_shoot        BOOLEAN      NOT NULL DEFAULT FALSE,
+    outdoor_contact_person  VARCHAR(100),
+    outdoor_contact_info    VARCHAR(200),
+    -- 线索
+    lead_source             VARCHAR(300),
+    -- 责编备注
+    notes                   TEXT,
+    created_at              TIMESTAMPTZ  NOT NULL DEFAULT NOW(),
+    updated_at              TIMESTAMPTZ  NOT NULL DEFAULT NOW()
+);
+
+CREATE INDEX idx_episodes_air_date     ON episodes (air_date);
+CREATE INDEX idx_episodes_editor_id    ON episodes (editor_id);
+CREATE INDEX idx_episodes_is_rerun     ON episodes (is_rerun);
+
+-- ============================================================
+-- 3. yearly_targets — 年度收视目标（Phase 2 用，只增不改）
+-- ============================================================
+CREATE TABLE IF NOT EXISTS yearly_targets (
+    id              SERIAL PRIMARY KEY,
+    year            INTEGER       NOT NULL UNIQUE,
+    base_target     NUMERIC(6,4)  NOT NULL,
+    stretch_target  NUMERIC(6,4)  NOT NULL,
+    issued_date     DATE,
+    notes           TEXT,
+    created_at      TIMESTAMPTZ   NOT NULL DEFAULT NOW()
+);
+
+-- ============================================================
+-- 4. topics — 选题 / 报题单（Phase 4a 用）
+-- ============================================================
+CREATE TABLE IF NOT EXISTS topics (
+    id                  SERIAL PRIMARY KEY,
+    title               VARCHAR(200) NOT NULL,
+    -- 内容概要（双段结构: 引子 + 列举）
+    content_summary     TEXT,
+    -- 编导关联（软引用 + 快照）
+    editor_id           INTEGER      REFERENCES users(id) ON DELETE SET NULL,
+    editor_name_snapshot VARCHAR(100),
+    -- 选题状态: draft=草稿, submitted=已提交, reviewing=审阅中, approved=通过, rejected=退回
+    status              VARCHAR(20)  NOT NULL DEFAULT 'draft'
+                        CHECK (status IN ('draft', 'submitted', 'reviewing', 'approved', 'rejected')),
+    -- AI 反馈（JSON 格式，含查重/画像/知识库参考等模块结果）
+    ai_feedback         JSONB,
+    -- 关联的 episodes / knowledge_items id 列表
+    related_episodes    JSONB,
+    related_knowledge   JSONB,
+    created_at          TIMESTAMPTZ  NOT NULL DEFAULT NOW(),
+    updated_at          TIMESTAMPTZ  NOT NULL DEFAULT NOW()
+);
+
+CREATE INDEX idx_topics_editor_id ON topics (editor_id);
+CREATE INDEX idx_topics_status    ON topics (status);
+
+-- ============================================================
+-- 5. topic_embeddings — 选题语义向量（Phase 4a 用，pgvector）
+-- ============================================================
+CREATE TABLE IF NOT EXISTS topic_embeddings (
+    id          SERIAL PRIMARY KEY,
+    topic_id    INTEGER       NOT NULL REFERENCES topics(id) ON DELETE CASCADE,
+    embedding   vector(1536)  NOT NULL,
+    created_at  TIMESTAMPTZ   NOT NULL DEFAULT NOW()
+);
+
+CREATE INDEX idx_topic_embeddings_topic_id ON topic_embeddings (topic_id);
+-- pgvector 向量索引（IVFFlat，数据量上去后可用 HNSW）
+-- CREATE INDEX idx_topic_embeddings_vector ON topic_embeddings USING ivfflat (embedding vector_cosine_ops);
+
+-- ============================================================
+-- 6. schedules — 排期（Phase 4b 用）
+-- ============================================================
+CREATE TABLE IF NOT EXISTS schedules (
+    id                  SERIAL PRIMARY KEY,
+    episode_id          INTEGER      REFERENCES episodes(id) ON DELETE SET NULL,
+    planned_air_date    DATE         NOT NULL,
+    status              VARCHAR(20)  NOT NULL DEFAULT 'planned'
+                        CHECK (status IN ('planned', 'confirmed', 'completed', 'cancelled')),
+    editor_id           INTEGER      REFERENCES users(id) ON DELETE SET NULL,
+    notes               TEXT,
+    created_at          TIMESTAMPTZ  NOT NULL DEFAULT NOW(),
+    updated_at          TIMESTAMPTZ  NOT NULL DEFAULT NOW()
+);
+
+CREATE INDEX idx_schedules_air_date ON schedules (planned_air_date);
+
+-- ============================================================
+-- 7. knowledge_items — 知识库条目（Phase 3 用，2.0 兼容预埋）
+-- ============================================================
+CREATE TABLE IF NOT EXISTS knowledge_items (
+    id                  SERIAL PRIMARY KEY,
+    title               VARCHAR(300) NOT NULL,
+    -- Markdown 格式正文（向 2.0 Karpathy LLM Wiki 形态靠拢）
+    content_md          TEXT,
+    -- 来源信息
+    source_type         VARCHAR(30)  NOT NULL DEFAULT 'manual'
+                        CHECK (source_type IN ('baoti', 'manuscript', 'military_report', 'manual', 'rss')),
+    source_file_name    VARCHAR(300),
+    source_url          VARCHAR(1000),
+    -- 作者 / 编导
+    author              VARCHAR(100),
+    publish_date        DATE,
+    -- ⚠️ 2.0 兼容预埋字段（1.0 不用但要建）
+    tags                JSONB        DEFAULT '[]'::jsonb,
+    related_entities    JSONB        DEFAULT '[]'::jsonb,
+    related_concepts    JSONB        DEFAULT '[]'::jsonb,
+    created_at          TIMESTAMPTZ  NOT NULL DEFAULT NOW(),
+    updated_at          TIMESTAMPTZ  NOT NULL DEFAULT NOW()
+);
+
+CREATE INDEX idx_knowledge_items_source_type ON knowledge_items (source_type);
+CREATE INDEX idx_knowledge_items_tags ON knowledge_items USING gin (tags);
+
+-- ============================================================
+-- 8. knowledge_embeddings — 知识库向量（Phase 3 用，pgvector）
+-- ============================================================
+CREATE TABLE IF NOT EXISTS knowledge_embeddings (
+    id              SERIAL PRIMARY KEY,
+    knowledge_id    INTEGER       NOT NULL REFERENCES knowledge_items(id) ON DELETE CASCADE,
+    chunk_index     INTEGER       NOT NULL DEFAULT 0,
+    chunk_text      TEXT          NOT NULL,
+    embedding       vector(1536)  NOT NULL,
+    created_at      TIMESTAMPTZ   NOT NULL DEFAULT NOW()
+);
+
+CREATE INDEX idx_knowledge_embeddings_knowledge_id ON knowledge_embeddings (knowledge_id);
+-- pgvector 向量索引
+-- CREATE INDEX idx_knowledge_embeddings_vector ON knowledge_embeddings USING ivfflat (embedding vector_cosine_ops);
+
+-- ============================================================
+-- 9. review_comments — 审片意见（Phase 4b 用）
+-- ============================================================
+CREATE TABLE IF NOT EXISTS review_comments (
+    id              SERIAL PRIMARY KEY,
+    episode_id      INTEGER       NOT NULL REFERENCES episodes(id) ON DELETE CASCADE,
+    comment_text    TEXT          NOT NULL,
+    reviewer        VARCHAR(100),
+    created_at      TIMESTAMPTZ   NOT NULL DEFAULT NOW()
+);
+
+CREATE INDEX idx_review_comments_episode_id ON review_comments (episode_id);
+
+-- ============================================================
+-- 10. shooting_resources — 拍摄资源 / 人脉（Phase 4b 用）
+-- ============================================================
+CREATE TABLE IF NOT EXISTS shooting_resources (
+    id              SERIAL PRIMARY KEY,
+    contact_person  VARCHAR(100) NOT NULL,
+    contact_info    VARCHAR(200),
+    organization    VARCHAR(200),
+    -- 资源类型: personnel=人员, location=场地, equipment=设备, other=其他
+    resource_type   VARCHAR(20)  DEFAULT 'other'
+                    CHECK (resource_type IN ('personnel', 'location', 'equipment', 'other')),
+    notes           TEXT,
+    created_at      TIMESTAMPTZ  NOT NULL DEFAULT NOW(),
+    updated_at      TIMESTAMPTZ  NOT NULL DEFAULT NOW()
+);
+
+-- ============================================================
+-- 11. radar_items — 热点雷达条目（Phase 4c 用）
+-- ============================================================
+CREATE TABLE IF NOT EXISTS radar_items (
+    id              SERIAL PRIMARY KEY,
+    title           VARCHAR(300) NOT NULL,
+    source_url      VARCHAR(1000),
+    source_name     VARCHAR(100),
+    summary         TEXT,
+    -- AI 处理结果
+    ai_summary      TEXT,
+    style_summary   TEXT,
+    published_at    TIMESTAMPTZ,
+    fetched_at      TIMESTAMPTZ  NOT NULL DEFAULT NOW(),
+    created_at      TIMESTAMPTZ  NOT NULL DEFAULT NOW()
+);
+
+-- ============================================================
+-- 12. radar_user_actions — 编导对热点的操作（Phase 4c 用）
+-- ============================================================
+CREATE TABLE IF NOT EXISTS radar_user_actions (
+    id              SERIAL PRIMARY KEY,
+    radar_item_id   INTEGER       NOT NULL REFERENCES radar_items(id) ON DELETE CASCADE,
+    user_id         INTEGER       NOT NULL REFERENCES users(id) ON DELETE CASCADE,
+    -- 操作类型: deep_dive=深入了解, bookmark=收藏, ignore=忽略
+    action_type     VARCHAR(20)   NOT NULL
+                    CHECK (action_type IN ('deep_dive', 'bookmark', 'ignore')),
+    created_at      TIMESTAMPTZ   NOT NULL DEFAULT NOW(),
+    UNIQUE (radar_item_id, user_id, action_type)
+);
+
+-- ============================================================
+-- 初始化完成
+-- ============================================================

backend/sql/002_add_cover_settings.sql

@@ -0,0 +1,23 @@
+-- ============================================================
+-- 002_add_cover_settings.sql — 题图设置表
+-- 执行方式: psql -U postgres -d milsci_dev -f 002_add_cover_settings.sql
+-- ============================================================
+
+-- 题图设置表（只存当前展示的一张题图，不动 episodes 表）
+CREATE TABLE IF NOT EXISTS cover_settings (
+    id              SERIAL PRIMARY KEY,
+    key             VARCHAR(50) NOT NULL UNIQUE DEFAULT 'dashboard_cover',
+    cover_path      TEXT,                            -- 当前题图文件路径（相对路径，如 /static/covers/xxx.jpg）
+    episode_number  INTEGER,                          -- 关联期次号（可空）
+    episode_title   TEXT,                            -- 关联期次节目名（可空）
+    updated_at      TIMESTAMPTZ NOT NULL DEFAULT NOW()
+);
+
+-- 初始化默认值行（允许为空，表示使用默认渐变底图）
+INSERT INTO cover_settings (key, cover_path, episode_number, episode_title)
+VALUES ('dashboard_cover', NULL, NULL, NULL)
+ON CONFLICT (key) DO NOTHING;
+
+-- ============================================================
+-- 迁移完成
+-- ============================================================

backend/sql/002a_diagnostic_episode_number.sql

@@ -0,0 +1,31 @@
+-- ============================================================
+-- 002a_diagnostic_episode_number.sql
+-- 诊断脚本：将 episodes.episode_number 从全局 UNIQUE 改为 (air_year, episode_number) 复合 UNIQUE
+-- 执行方式: psql -U postgres -d milsci_dev -f 002a_diagnostic_episode_number.sql
+-- 前置条件：先设 $env:PGCLIENTENCODING="UTF8"
+-- ============================================================
+
+-- 三项诊断（任一异常停下找制片人）
+SELECT
+  'total_rows'           as check_name,
+  count(*)               as value
+FROM episodes
+UNION ALL
+SELECT
+  'null_air_date'        as check_name,
+  count(*)               as value
+FROM episodes
+WHERE air_date IS NULL
+UNION ALL
+SELECT
+  'duplicate_year_epnum' as check_name,
+  count(*)               as value
+FROM (
+  SELECT
+    EXTRACT(YEAR FROM air_date)::int as air_year,
+    episode_number
+  FROM episodes
+  WHERE original_episode_id IS NULL
+  GROUP BY 1, 2
+  HAVING count(*) > 1
+) as dup;

backend/sql/002b_migrate_episode_number.sql

@@ -0,0 +1,18 @@
+-- ============================================================
+-- 002b_migrate_episode_number.sql
+-- 迁移脚本：在 002a 诊断确认安全后执行
+-- 执行方式: psql -U postgres -d milsci_dev -f sql/002b_migrate_episode_number.sql
+-- 前置条件：先设 $env:PGCLIENTENCODING="UTF8"
+-- 注意：已改用 date_part('year', air_date) 替代 EXTRACT(... )::int 避免语法错误
+-- ============================================================
+
+-- 1. 删全局 UNIQUE 约束（幂等：constraint 已删则报 warning，正常）
+ALTER TABLE episodes DROP CONSTRAINT episodes_episode_number_key;
+
+-- 2. 建按年复合唯一索引（重播行 original_episode_id IS NOT NULL 不参与）
+CREATE UNIQUE INDEX idx_episodes_year_number
+  ON episodes (date_part('year', air_date), episode_number)
+  WHERE original_episode_id IS NULL;
+
+-- 验证：确认索引生效
+-- \d episodes

backend/sql/002c_recreate_index_only.sql

@@ -0,0 +1,10 @@
+-- ============================================================
+-- 002c_recreate_index_only.sql
+-- 002b 跑半截翻车后补救：只执行 CREATE INDEX，不要再 DROP CONSTRAINT
+-- 执行方式: psql -U postgres -d milsci_dev -f sql/002c_recreate_index_only.sql
+-- 前置条件：先设 $env:PGCLIENTENCODING="UTF8"
+-- ============================================================
+
+CREATE UNIQUE INDEX idx_episodes_year_number
+  ON episodes (date_part('year', air_date), episode_number)
+  WHERE original_episode_id IS NULL;

claude协作原则.md

@@ -0,0 +1,13 @@
+**本文件是制片人刘通与 Claude(Opus 4.7 高级技术顾问)的协作原则,与项目宪法(.clinerules / project_plan.md / dev_plan.md / docs/git_workflow.md)并列。新对话接手时,请连同宪法一并阅读本文件。**
+
+---
+
+**Claude 与我交流、沟通、协作时需要注意的原则**
+
+1. 全程使用简体中文。
+
+2. 我不懂代码编程,给我的解释说明和理由要尽量简明扼要。不要堆叠大量技术语言,说人话。
+
+3. 请参阅宪法文件中你(Claude)的定位:写具体 Plan 是 Cline 的事,不要架空 Cline 的 Plan 模式。除非特别必要(Cline 反复修改都过不去的坎)可以上手,以及制片人明确指定你可以操刀的具体工作。
+
+4. 给 Cline 的指令要用代码块封装起来,方便复制。

dev_plan.md

@@ -2,7 +2,7 @@
 
 > 项目:TPS(Topic Planning System)中台
 > 仓库:`tps-dashboard`(Gitea)
-> 修订版本:v3 · 2026-05-06
+> 修订版本:v5 · 2026-05-27
 > 配套文档:`project_plan.md`(产品宪法)+ `.clinerules`(协作规则)+ `docs/git_workflow.md`(Git 操作手册)
 
 ---
@@ -62,7 +62,7 @@
 **路径 B — 迁移(推荐)**:
 1. 制片人在家用机把 `backend/` 整个目录打包 zip
 2. 拷到单位机,解压进 `tps-dashboard/backend/`
-3. Cline 跑一遍 Plan,Sonnet 审一遍代码确认与新版宪法一致
+3. Cline 跑一遍 Plan,MiniMax M2.7 审一遍代码确认与新版宪法一致
 4. 修任何不一致(目前看应无重大不一致)
 5. 跑一遍 `pip install + psql + create_admin + uvicorn` 验证
 
@@ -90,8 +90,7 @@
 | 向量检索 | **pgvector**(PG 扩展) | 不用独立向量库 |
 | 向量维度 | **1536**(为 2.0 wiki 兼容) | 不可改 |
 | AI 嵌入模型 | DeepSeek 或 OpenAI 兼容 embedding API | — |
-| AI 编程模型 | Phase 0-1: Sonnet Plan + DeepSeek pro Act | 详见 `.clinerules` 第二章 |
-| AI 编程模型 | Phase 2 起: MiniMax M2.7 Plan + DeepSeek pro Act | — |
+| AI 编程模型 | **MiniMax M2.7**(Cline Plan + Act 模式) | 详见 `.clinerules` 第二章 |
 | 认证 | **Session + Cookie** | 不用 JWT |
 | Word 生成 | docxtpl | — |
 | 甘特图前端库 | frappe-gantt | 不自己用 D3 画 |
@@ -142,7 +141,7 @@
 **Day 2:开发工具**
 
 - VS Code + Cline 扩展
-- Cline 配置 Plan 模式(Sonnet)+ Act 模式(DeepSeek pro)
+- Cline 配置 Plan 模式(MiniMax M2.7)+ Act 模式(MiniMax M2.7),Plan 和 Act 用同一模型但物理隔离两个心智阶段(纪律性)
 - VS Code 终端打开测试:能跑 `python --version` / `node --version` / `git --version` / `psql --version`
 
 **Phase 0 收尾**:
@@ -163,7 +162,7 @@ git push origin phase0-env-ready
 
 **Day 3-4:后端骨架**
 
-- **路径 B 推荐**:从家用机迁移 backend 整目录,Sonnet 审一遍,跑通验证
+- **路径 B 推荐**:从家用机迁移 backend 整目录,MiniMax M2.7 审一遍,跑通验证
 - **路径 A 备选**:从零开始重写
 - 验收:`uvicorn app.main:app --reload` 能起,`/docs` 页面能开,`/api/auth/login` + `/api/auth/me` + `/api/auth/logout` 三接口验证通过
 
@@ -202,7 +201,7 @@ git push origin phase0-env-ready
 - 编辑 episodes / yearly_targets / users 的 SQLModel 模型
 
 **前端**:
-- 仪表盘页面(近 4 期收视 + 未来 6 期排期)
+- 仪表盘页面(近 9 期收视 + 未来 6 期排期)
 - 责编录入界面
 - 管理员用户管理界面
 
@@ -213,7 +212,7 @@ git push origin phase0-env-ready
 - 后端用 pandas 解析,批量 INSERT
 - **死规矩**:导入时检测 yearly_targets 是否对应年份齐全,缺则报错让责编先填(防止颜色判定时找不到目标)
 
-**Phase 2 起 Plan 模式切换为 MiniMax M2.7**:Sonnet 余额耗尽前最后一个由 Sonnet 出方案的 Phase,Phase 2 末做切换的复盘评估。
+**Phase 2 已进入**:Plan = Act = MiniMax M2.7,DeepSeek v4 pro 保留为备选。
 
 **Phase 2 里程碑**:
 - 责编可以开始录入(或导入)历史收视数据
@@ -250,19 +249,23 @@ git push origin phase0-env-ready
 
 ---
 
-### Phase 4a:核心 MVP(Week 5-6)
+### Phase 4a:核心 MVP — 对话式 TPS(Week 5-6)
 
-**模块 G**:报题单生成与下载(docxtpl 模板)
-- 模板严格遵守"双段结构"(`.clinerules` 5.4)
+> v5 升级:模块 A/C/G 合成"对话式策划助手"形态(左对话右报告、脚注式引用),详见 `project_plan.md` 模块 A/C/G + 第 2.3 节 + 6.5 节红线。
 
-**模块 A**:选题查重(语义检索 + 历史收视展示)
-- 联动模块 E 展示外拍资源
+**前置条件(硬门槛,不满足不开工)**:
+- 知识库笔记已批量录入(Backlog#2),收视数据已真实导入(Phase 2 遗留债)。**库空时对话助手"无米下锅",依据全空会毁掉第一印象。**
+- Phase 3 的**朴素语义检索**已先行跑通(输入文本→返回最相关条目,带相关度/原文片段/来源),作为对话式报告的检索底座。
 
-**模块 C**:知识库参考(检索往期文稿 / 军报)
+**模块 A(对话式策划主入口)**:左栏编导对话、右栏 AI 加工的策划参考报告(历史是否做过/收视/审片意见/可参考节目/外拍资源/知识库依据)。严守三条交互红线:求助式不投喂、脚注式锚定引用、持续对话沉淀编导判断(见 `project_plan.md` 模块 A)。
 
-**前端**:选题提交表单 + AI 反馈页面布局
+**模块 C(知识库依据供给)**:检索结果融入右栏报告作脚注依据,而非孤立列表。
 
-**🎯 Week 6 末上线 MVP**——团队真的可以用了:登录、看仪表盘、提交选题、看到查重和知识库结果、生成报题单 docx 下载。
+**模块 G(报题单生成)**:聊定后一键按范本规范化生成 docx。**只做规范化呈现,不代写立意**(见模块 G 分寸线 + 6.5 红线)。模板严守"双段结构"(`.clinerules` 5.4)。
+
+**AI 成本机制**:本 Phase 起对话调用大模型,按第十一章"AI 成本与额度原则"落地——按 token 记、按人+选题归因、保守上限 + 超额提醒管理员。**朴素的一张用量表即可,严禁上 AI 网关/可观测性框架等大公司重型武器。**
+
+**🎯 Week 6 末上线 MVP**——团队真的可以用了:登录、看仪表盘、与 TPS 对话策划选题、看到有依据的策划报告、聊定后一键生成报题单 docx 下载。
 
 **收尾**:写 `logs/phase4a_log.md`,打 `phase4a-mvp-launch` tag。**这是项目第一次"对内发布",务必郑重打 tag**。
 
@@ -486,7 +489,9 @@ tps-dashboard/                  ← Git 仓库根
 |---|---|---|
 | v1 | 2026-04 | 初版 |
 | v2 | 2026-05-04 | 精简技术栈、Phase 拆分(避免 Phase 4 大爆炸)、明确每周里程碑 |
-| **v3** | **2026-05-06** | **加 Git + Gitea 工作流;Phase 0 重排(纳入 Git);Phase 2 起 Plan 切 MiniMax M2.7;明确单位机重启路径(路径 B 推荐);新增 docs/ 目录与 phase{N} tag 收尾仪式** |
+| v3 | 2026-05-06 | 加 Git + Gitea 工作流;Phase 0 重排(纳入 Git);Phase 2 起 Plan 切 MiniMax M2.7;明确单位机重启路径(路径 B 推荐);新增 docs/ 目录与 phase{N} tag 收尾仪式 |
+| v4 | 2026-05-14 | 家用机退役,唯一开发环境为单位 4090D;Phase 2 起 MiniMax M2.7 正式日常使用;更新版本演进记录 |
+| **v5** | **2026-05-27** | **Phase 4a 升级为"对话式 TPS"(对齐 project_plan v5):模块 A/C/G 合成左对话右报告形态;补硬门槛(数据灌入 + Phase 3 朴素检索先行);加 AI 成本机制落地说明** |
 
 ---
 

docs/backlog_md_TPS_工作台待办池_不丢视野_非排期_.md

@@ -0,0 +1,88 @@
+# backlog.md — TPS 工作台待办池（不丢视野，非排期）
+
+> 维护人：刘通 + 顾问 Opus
+> 用途：记录"暂不做但不能漏"的事项。区别于 phase 计划，这里是池子，
+>    每个 Phase 收尾时回看一遍，决定捞哪条进下个 Phase。
+
+---
+
+## A. 定位澄清（认知，非待办）
+
+- 本系统对外称"中台/工作台"，实质是**单栏目业务系统**（一条业务线、8 用户）。
+- 不套真中台那套（多租户/服务网格/配置中心/跨线统一权限）——对 8 人是纯负担。
+- 真正定位：把散在 Excel/Word/微信/个人电脑/师徒口传的工作流，收拢成
+  有沉淀、有权限、可检索的系统。核心价值 = 经验从"靠人"变"靠系统"。
+- 现有克制选择（不用 TS / 不用 Alembic / pgvector 不用独立向量库 /
+  Session 不用 JWT）方向正确，保持。
+
+---
+
+## B. 工作台核心能力缺口（按优先级）
+
+### B1. 选题状态流转【优先级：高】
+- 现状：episodes 存收视+元数据，但"提报→立项→排期→拍摄→制作→播出→复盘"
+  这条状态线是散的。
+- 建议：给 topics/episodes 加状态机字段，制片人一眼看到每个选题卡在哪环。
+- 与模块 F 甘特图区别：甘特是排期视角，这是业务状态视角，两回事。
+- 触发时机：Phase 4b（TPS 闭环时一并补最自然）。
+
+### B2. 操作留痕 / 审计【优先级：高，成本极低，建议早埋】
+- 现状：谁改收视、谁删选题、谁换题图，全靠信任，无记录。
+- 建议：核心业务表（episodes/topics/yearly_targets）加
+  created_by / updated_by / updated_at 字段即可，不必做复杂日志系统。
+- 触发时机：越早埋越好，建议 Phase 3 起手时顺带加（改表成本最低）。
+
+### B3. 数据导出 / 备份【优先级：中】
+- 现状：Phase 2 做了 Excel 导入，导出是它的镜像，迟早要。
+- 收视、选题是栏目资产，将来要能导出 Excel、能定期备份。
+- 触发时机：Phase 5 上线前后。
+
+### B4. 通知 / 待办提醒【优先级：低】
+- "选题被审了""该交稿了"这类。小团队可能微信/Mattermost 群就解决了。
+
+- 若 B 见下方与 Mattermost 打通，可借 IM 发通知，不必自建。
+
+- 触发时机：看实际使用反馈再定，可能不做。
+
+  ### B5.关联【优先级：低】
+
+  题图"自动算近5期最高份额提示"本期做的是静态文案占位,  自动计算挪二期(原计划)。
+
+---
+
+## C. 跨系统整合（与现有其他系统合服务器、打通）
+
+### C1. 蓝皓配音 TTS 2.0 合服务器【待商议】
+- 背景：TTS 2.0 现为独立网站/服务器，单租服务器费钱。
+- 设想：与 TPS 中台挪到同一台服务器，省一份租金。
+- 待办：制片人把 TTS 2.0 源代码给 Opus → 一起评估
+  ①技术栈是否冲突（端口/依赖/数据库）②Nginx 反代能否共存
+  ③资源够不够（4核8G 轻量服务器跑两套+PG+pgvector 要算内存账）。
+- 注：dev_plan 部署章已写"Nginx 反代整合 TPS / 蓝皓 TTS / MMIM"，本条是落实。
+- 触发时机：Phase 5 部署规划时。
+
+### C2. Mattermost(内部 IM) 打通【待商议】
+- 背景：团队内部 IM 用 Mattermost，部署在与 TTS 2.0 同一服务器。
+- 设想：TPS 的通知/待办（见 B4）可推到 Mattermost，不必自建通知系统。
+- 待办：评估 Mattermost incoming webhook 接入成本（通常很低）。
+- 触发时机：B4 真要做时一并考虑。
+
+---
+
+## D. 已知技术隐患（从 phase log 汇集，不漏）
+
+- backend/ 补 venv（全局依赖搬进去）。
+- 隐患 A：ImportError 与 Python 内置同名（低优）。
+- 隐患 B：ExcelService 返回 Pydantic 化（低优）。
+- Phase 5 部署安全：SECRET_KEY 换真随机、same_site=strict + https_only=True、
+  上 SSL、关 /docs。
+- Phase 5 上线前数据清理：清测试数据（期次 8888/8889、年度目标 2027/2028
+  占位等），录真数据。
+- logo 路径用 /src/assets/...(Vite开发模式),生产打包会失效,Phase5部署改。 - reset_password.py 能改任意用户密码,Phase5部署前不能暴露在生产环境。 - 柱图设计为取最近12期,当前真实数据仅7期故显7根;Phase5录真数据后自动补齐,届时确认满12根正常。 - 完成率用 new Date().getFullYear() 取当前年:每年初需录入当年 yearly_targets,  否则跨年后两项完成率显"—"。(运营提醒,非bug)
+
+---
+
+## E. 明确不做（防止被"中台"概念诱导加功能）
+
+- 多租户、服务网格、配置中心、跨业务线统一权限平台 —— 8 人单栏目纯负担。
+- 任何让 AI 替编导拍板的功能 —— 违背"AI 给方向不给答案"。

docs/git_workflow.md

@@ -2,7 +2,7 @@
 
 > 项目:TPS 中台
 > 仓库:`tps-dashboard`(单位 Gitea)
-> 文档版本:v1 · 2026-05-06
+> 文档版本:v2 · 2026-05-14
 > 受众:**制片人(不会 Git)** + Cline(自动遵守此规则)
 
 ---
@@ -22,7 +22,7 @@
 
 ### 1.1 三个核心概念,一分钟搞定
 
-**Git 仓库**:就是一个"被 Git 管起来的文件夹"。打开本地 `D:\TPS-中台\` 这个文件夹,里面有所有项目文件 + 一个隐藏的 `.git` 子目录。**不要删 `.git`**,删了等于把保险撕了。
+**Git 仓库**:就是一个"被 Git 管起来的文件夹"。打开本地 `E:\tps-dashboard\` 这个文件夹,里面有所有项目文件 + 一个隐藏的 `.git` 子目录。**不要删 `.git`**,删了等于把保险撕了。
 
 **commit(提交)**:相当于"存档"。每一次 commit 是一个时间点的快照,以后可以回到这个时间点。
 
@@ -71,7 +71,7 @@ git clone http://localhost:3000/simonkoson/tps-dashboard.git
 
 ```powershell
 # 切换到你想放项目的位置
-cd Z:\AIworks\
+cd E:\tps-dashboard\
 
 # 把刚才复制的 clone 命令粘贴运行
 git clone http://localhost:3000/simonkoson/tps-dashboard.git
@@ -88,12 +88,12 @@ done.
 
 `warning: empty repository` 是**正常的**——本来就是空仓库,不是错误。
 
-完成后 `Z:\AIworks\` 下多了一个 `tps-dashboard` 文件夹。
+完成后 `E:\tps-dashboard\` 下多了一个 `tps-dashboard` 文件夹。
 
 ### 2.3 进仓库目录
 
 ```powershell
-cd Z:\AIworks\tps-dashboard
+cd E:\tps-dashboard\
 ```
 
 之后所有 git 命令都在这个目录里跑。
@@ -104,17 +104,17 @@ cd Z:\AIworks\tps-dashboard
 
 **手动操作**(用 Windows 资源管理器):
 
-1. 在 `Z:\AIworks\tps-dashboard\` 下:
+1. 在 `E:\tps-dashboard\` 下:
    - 直接拷入 `.clinerules`、`project_plan.md`、`dev_plan.md`、`README.md`(平铺四份)
    
-2. 在 `Z:\AIworks\tps-dashboard\` 下**新建文件夹** `docs`,把 `git_workflow.md` 拷进去
+2. 在 `E:\tps-dashboard\` 下**新建文件夹** `docs`,把 `git_workflow.md` 拷进去
 
-3. 在 `Z:\AIworks\tps-dashboard\` 下**新建文件夹** `logs`,把 `phase1_log.md` 拷进去
+3. 在 `E:\tps-dashboard\` 下**新建文件夹** `logs`,把 `phase1_log.md` 拷进去
 
 **最终结构应该是**:
 
 ```
-Z:\AIworks\tps-dashboard\
+E:\tps-dashboard\
 ├── .clinerules
 ├── project_plan.md
 ├── dev_plan.md
@@ -132,7 +132,7 @@ Z:\AIworks\tps-dashboard\
 **操作**:在 VS Code 终端里执行:
 
 ```powershell
-cd Z:\AIworks\tps-dashboard
+cd E:\tps-dashboard\
 notepad .gitignore
 ```
 
@@ -244,7 +244,7 @@ branch 'main' set up to track 'origin/main'.
 ### 3.1 每天开工第一件事
 
 ```powershell
-cd Z:\AIworks\tps-dashboard
+cd E:\tps-dashboard\
 git status                # 看看本地有没有未提交的改动
 git pull                  # 拉 Gitea 最新(如果别的机器推过)
 ```
@@ -259,16 +259,16 @@ git status                # 看看有没有未提交的
 
 ### 3.3 让 Cline 帮你提交
 
-最方便的方式——**让 Sonnet 在 Plan 里下命令**:
+最方便的方式——**让 MiniMax M2.7 在 Plan 里下命令**:
 
 > "现在改完了,请帮我 commit + push,commit message 写: feat: 添加 XXX 功能"
 
-Cline(Sonnet → DeepSeek)会自动跑 git add / git commit / git push。**这是默认推荐方式**。
+Cline(MiniMax M2.7)会自动跑 git add / git commit / git push。**这是默认推荐方式**。
 
 ### 3.4 你自己手工提交(备选)
 
 ```powershell
-cd Z:\AIworks\tps-dashboard
+cd E:\tps-dashboard\
 git add -A
 git commit -m "fix: 修复登录页跳转问题"
 git push
@@ -386,7 +386,7 @@ git push origin phase1-complete
 ### Q5:换了一台机器,怎么把代码搞过来
 
 ```powershell
-cd Z:\AIworks\tps-dashboard
+cd E:\tps-dashboard\
 git clone http://localhost:3000/simonkoson/tps-dashboard.git
 cd tps-dashboard
 # 然后跟原来一样开工就行
@@ -404,9 +404,17 @@ git reset HEAD~1     # 撤回最近一次 commit,文件改动保留
 
 ---
 
-## 第 7 章:多设备同步(家用机 ↔ 单位机)
+## 第 7 章:多设备同步(已废弃)
 
-### 离开 A 机器前
+> ⚠️ **2026-05-14 更新**:家用机已于 2026-05-14 退役,项目统一在单位 4090D 开发。
+> 本章描述的多设备同步模式**已废弃**,保留仅供历史参考。
+> 当前工作流:单一开发机,每日开工 `git pull`,离开前 `git push`。
+
+---
+
+### 7.1 历史模式(已废弃)
+
+#### 离开 A 机器前
 
 ```powershell
 git status         # 看有没有未提交的
@@ -432,6 +440,7 @@ git status         # 确认干净
 | 版本 | 日期 | 变化 |
 |---|---|---|
 | v1 | 2026-05-06 | 初版 |
+| **v2** | **2026-05-14** | **项目目录从 NAS(Z:\AIworks\)迁移至本地 SSD(E:\tps-dashboard\);家用机退役,唯一开发环境为单位 4090D;Phase 2 起 MiniMax M2.7 正式日常使用;更新版本演进记录** |
 
 ---
 

frontend/.gitignore

@@ -0,0 +1,24 @@
+# Logs
+logs
+*.log
+npm-debug.log*
+yarn-debug.log*
+yarn-error.log*
+pnpm-debug.log*
+lerna-debug.log*
+
+node_modules
+dist
+dist-ssr
+*.local
+
+# Editor directories and files
+.vscode/*
+!.vscode/extensions.json
+.idea
+.DS_Store
+*.suo
+*.ntvs*
+*.njsproj
+*.sln
+*.sw?

frontend/README.md

@@ -0,0 +1,16 @@
+# React + Vite
+
+This template provides a minimal setup to get React working in Vite with HMR and some ESLint rules.
+
+Currently, two official plugins are available:
+
+- [@vitejs/plugin-react](https://github.com/vitejs/vite-plugin-react/blob/main/packages/plugin-react) uses [Oxc](https://oxc.rs)
+- [@vitejs/plugin-react-swc](https://github.com/vitejs/vite-plugin-react/blob/main/packages/plugin-react-swc) uses [SWC](https://swc.rs/)
+
+## React Compiler
+
+The React Compiler is not enabled on this template because of its impact on dev & build performances. To add it, see [this documentation](https://react.dev/learn/react-compiler/installation).
+
+## Expanding the ESLint configuration
+
+If you are developing a production application, we recommend using TypeScript with type-aware lint rules enabled. Check out the [TS template](https://github.com/vitejs/vite/tree/main/packages/create-vite/template-react-ts) for information on how to integrate TypeScript and [`typescript-eslint`](https://typescript-eslint.io) in your project.

frontend/eslint.config.js

@@ -0,0 +1,21 @@
+import js from '@eslint/js'
+import globals from 'globals'
+import reactHooks from 'eslint-plugin-react-hooks'
+import reactRefresh from 'eslint-plugin-react-refresh'
+import { defineConfig, globalIgnores } from 'eslint/config'
+
+export default defineConfig([
+  globalIgnores(['dist']),
+  {
+    files: ['**/*.{js,jsx}'],
+    extends: [
+      js.configs.recommended,
+      reactHooks.configs.flat.recommended,
+      reactRefresh.configs.vite,
+    ],
+    languageOptions: {
+      globals: globals.browser,
+      parserOptions: { ecmaFeatures: { jsx: true } },
+    },
+  },
+])

frontend/index.html

@@ -0,0 +1,13 @@
+<!doctype html>
+<html lang="en">
+  <head>
+    <meta charset="UTF-8" />
+    <link rel="icon" type="image/svg+xml" href="/favicon.svg" />
+    <meta name="viewport" content="width=device-width, initial-scale=1.0" />
+    <title>frontend</title>
+  </head>
+  <body>
+    <div id="root"></div>
+    <script type="module" src="/src/main.jsx"></script>
+  </body>
+</html>

frontend/package.json

@@ -0,0 +1,33 @@
+{
+  "name": "frontend",
+  "private": true,
+  "version": "0.0.0",
+  "type": "module",
+  "scripts": {
+    "dev": "vite",
+    "build": "vite build",
+    "lint": "eslint .",
+    "preview": "vite preview"
+  },
+  "dependencies": {
+    "@ant-design/icons": "^6.2.3",
+    "antd": "^6.4.2",
+    "axios": "^1.16.1",
+    "dayjs": "^1.11.20",
+    "react": "^19.2.6",
+    "react-dom": "^19.2.6",
+    "react-router-dom": "^7.15.1",
+    "zustand": "^5.0.13"
+  },
+  "devDependencies": {
+    "@eslint/js": "^10.0.1",
+    "@types/react": "^19.2.14",
+    "@types/react-dom": "^19.2.3",
+    "@vitejs/plugin-react": "^6.0.1",
+    "eslint": "^10.3.0",
+    "eslint-plugin-react-hooks": "^7.1.1",
+    "eslint-plugin-react-refresh": "^0.5.2",
+    "globals": "^17.6.0",
+    "vite": "^8.0.12"
+  }
+}

frontend/public/icons.svg

@@ -0,0 +1,24 @@
+<svg xmlns="http://www.w3.org/2000/svg">
+  <symbol id="bluesky-icon" viewBox="0 0 16 17">
+    <g clip-path="url(#bluesky-clip)"><path fill="#08060d" d="M7.75 7.735c-.693-1.348-2.58-3.86-4.334-5.097-1.68-1.187-2.32-.981-2.74-.79C.188 2.065.1 2.812.1 3.251s.241 3.602.398 4.13c.52 1.744 2.367 2.333 4.07 2.145-2.495.37-4.71 1.278-1.805 4.512 3.196 3.309 4.38-.71 4.987-2.746.608 2.036 1.307 5.91 4.93 2.746 2.72-2.746.747-4.143-1.747-4.512 1.702.189 3.55-.4 4.07-2.145.156-.528.397-3.691.397-4.13s-.088-1.186-.575-1.406c-.42-.19-1.06-.395-2.741.79-1.755 1.24-3.64 3.752-4.334 5.099"/></g>
+    <defs><clipPath id="bluesky-clip"><path fill="#fff" d="M.1.85h15.3v15.3H.1z"/></clipPath></defs>
+  </symbol>
+  <symbol id="discord-icon" viewBox="0 0 20 19">
+    <path fill="#08060d" d="M16.224 3.768a14.5 14.5 0 0 0-3.67-1.153c-.158.286-.343.67-.47.976a13.5 13.5 0 0 0-4.067 0c-.128-.306-.317-.69-.476-.976A14.4 14.4 0 0 0 3.868 3.77C1.546 7.28.916 10.703 1.231 14.077a14.7 14.7 0 0 0 4.5 2.306q.545-.748.965-1.587a9.5 9.5 0 0 1-1.518-.74q.191-.14.372-.293c2.927 1.369 6.107 1.369 8.999 0q.183.152.372.294-.723.437-1.52.74.418.838.963 1.588a14.6 14.6 0 0 0 4.504-2.308c.37-3.911-.63-7.302-2.644-10.309m-9.13 8.234c-.878 0-1.599-.82-1.599-1.82 0-.998.705-1.82 1.6-1.82.894 0 1.614.82 1.599 1.82.001 1-.705 1.82-1.6 1.82m5.91 0c-.878 0-1.599-.82-1.599-1.82 0-.998.705-1.82 1.6-1.82.893 0 1.614.82 1.599 1.82 0 1-.706 1.82-1.6 1.82"/>
+  </symbol>
+  <symbol id="documentation-icon" viewBox="0 0 21 20">
+    <path fill="none" stroke="#aa3bff" stroke-linecap="round" stroke-linejoin="round" stroke-width="1.35" d="m15.5 13.333 1.533 1.322c.645.555.967.833.967 1.178s-.322.623-.967 1.179L15.5 18.333m-3.333-5-1.534 1.322c-.644.555-.966.833-.966 1.178s.322.623.966 1.179l1.534 1.321"/>
+    <path fill="none" stroke="#aa3bff" stroke-linecap="round" stroke-linejoin="round" stroke-width="1.35" d="M17.167 10.836v-4.32c0-1.41 0-2.117-.224-2.68-.359-.906-1.118-1.621-2.08-1.96-.599-.21-1.349-.21-2.848-.21-2.623 0-3.935 0-4.983.369-1.684.591-3.013 1.842-3.641 3.428C3 6.449 3 7.684 3 10.154v2.122c0 2.558 0 3.838.706 4.726q.306.383.713.671c.76.536 1.79.64 3.581.66"/>
+    <path fill="none" stroke="#aa3bff" stroke-linecap="round" stroke-linejoin="round" stroke-width="1.35" d="M3 10a2.78 2.78 0 0 1 2.778-2.778c.555 0 1.209.097 1.748-.047.48-.129.854-.503.982-.982.145-.54.048-1.194.048-1.749a2.78 2.78 0 0 1 2.777-2.777"/>
+  </symbol>
+  <symbol id="github-icon" viewBox="0 0 19 19">
+    <path fill="#08060d" fill-rule="evenodd" d="M9.356 1.85C5.05 1.85 1.57 5.356 1.57 9.694a7.84 7.84 0 0 0 5.324 7.44c.387.079.528-.168.528-.376 0-.182-.013-.805-.013-1.454-2.165.467-2.616-.935-2.616-.935-.349-.91-.864-1.143-.864-1.143-.71-.48.051-.48.051-.48.787.051 1.2.805 1.2.805.695 1.194 1.817.857 2.268.649.064-.507.27-.857.49-1.052-1.728-.182-3.545-.857-3.545-3.87 0-.857.31-1.558.8-2.104-.078-.195-.349-1 .077-2.078 0 0 .657-.208 2.14.805a7.5 7.5 0 0 1 1.946-.26c.657 0 1.328.092 1.946.26 1.483-1.013 2.14-.805 2.14-.805.426 1.078.155 1.883.078 2.078.502.546.799 1.247.799 2.104 0 3.013-1.818 3.675-3.558 3.87.284.247.528.714.528 1.454 0 1.052-.012 1.896-.012 2.156 0 .208.142.455.528.377a7.84 7.84 0 0 0 5.324-7.441c.013-4.338-3.48-7.844-7.773-7.844" clip-rule="evenodd"/>
+  </symbol>
+  <symbol id="social-icon" viewBox="0 0 20 20">
+    <path fill="none" stroke="#aa3bff" stroke-linecap="round" stroke-linejoin="round" stroke-width="1.35" d="M12.5 6.667a4.167 4.167 0 1 0-8.334 0 4.167 4.167 0 0 0 8.334 0"/>
+    <path fill="none" stroke="#aa3bff" stroke-linecap="round" stroke-linejoin="round" stroke-width="1.35" d="M2.5 16.667a5.833 5.833 0 0 1 8.75-5.053m3.837.474.513 1.035c.07.144.257.282.414.309l.93.155c.596.1.736.536.307.965l-.723.73a.64.64 0 0 0-.152.531l.207.903c.164.715-.213.991-.84.618l-.872-.52a.63.63 0 0 0-.577 0l-.872.52c-.624.373-1.003.094-.84-.618l.207-.903a.64.64 0 0 0-.152-.532l-.723-.729c-.426-.43-.289-.864.306-.964l.93-.156a.64.64 0 0 0 .412-.31l.513-1.034c.28-.562.735-.562 1.012 0"/>
+  </symbol>
+  <symbol id="x-icon" viewBox="0 0 19 19">
+    <path fill="#08060d" fill-rule="evenodd" d="M1.893 1.98c.052.072 1.245 1.769 2.653 3.77l2.892 4.114c.183.261.333.48.333.486s-.068.089-.152.183l-.522.593-.765.867-3.597 4.087c-.375.426-.734.834-.798.905a1 1 0 0 0-.118.148c0 .01.236.017.664.017h.663l.729-.83c.4-.457.796-.906.879-.999a692 692 0 0 0 1.794-2.038c.034-.037.301-.34.594-.675l.551-.624.345-.392a7 7 0 0 1 .34-.374c.006 0 .93 1.306 2.052 2.903l2.084 2.965.045.063h2.275c1.87 0 2.273-.003 2.266-.021-.008-.02-1.098-1.572-3.894-5.547-2.013-2.862-2.28-3.246-2.273-3.266.008-.019.282-.332 2.085-2.38l2-2.274 1.567-1.782c.022-.028-.016-.03-.65-.03h-.674l-.3.342a871 871 0 0 1-1.782 2.025c-.067.075-.405.458-.75.852a100 100 0 0 1-.803.91c-.148.172-.299.344-.99 1.127-.304.343-.32.358-.345.327-.015-.019-.904-1.282-1.976-2.808L6.365 1.85H1.8zm1.782.91 8.078 11.294c.772 1.08 1.413 1.973 1.425 1.984.016.017.241.02 1.05.017l1.03-.004-2.694-3.766L7.796 5.75 5.722 2.852l-1.039-.004-1.039-.004z" clip-rule="evenodd"/>
+  </symbol>
+</svg>

frontend/src/App.css

@@ -0,0 +1,184 @@
+.counter {
+  font-size: 16px;
+  padding: 5px 10px;
+  border-radius: 5px;
+  color: var(--accent);
+  background: var(--accent-bg);
+  border: 2px solid transparent;
+  transition: border-color 0.3s;
+  margin-bottom: 24px;
+
+  &:hover {
+    border-color: var(--accent-border);
+  }
+  &:focus-visible {
+    outline: 2px solid var(--accent);
+    outline-offset: 2px;
+  }
+}
+
+.hero {
+  position: relative;
+
+  .base,
+  .framework,
+  .vite {
+    inset-inline: 0;
+    margin: 0 auto;
+  }
+
+  .base {
+    width: 170px;
+    position: relative;
+    z-index: 0;
+  }
+
+  .framework,
+  .vite {
+    position: absolute;
+  }
+
+  .framework {
+    z-index: 1;
+    top: 34px;
+    height: 28px;
+    transform: perspective(2000px) rotateZ(300deg) rotateX(44deg) rotateY(39deg)
+      scale(1.4);
+  }
+
+  .vite {
+    z-index: 0;
+    top: 107px;
+    height: 26px;
+    width: auto;
+    transform: perspective(2000px) rotateZ(300deg) rotateX(40deg) rotateY(39deg)
+      scale(0.8);
+  }
+}
+
+#center {
+  display: flex;
+  flex-direction: column;
+  gap: 25px;
+  place-content: center;
+  place-items: center;
+  flex-grow: 1;
+
+  @media (max-width: 1024px) {
+    padding: 32px 20px 24px;
+    gap: 18px;
+  }
+}
+
+#next-steps {
+  display: flex;
+  border-top: 1px solid var(--border);
+  text-align: left;
+
+  & > div {
+    flex: 1 1 0;
+    padding: 32px;
+    @media (max-width: 1024px) {
+      padding: 24px 20px;
+    }
+  }
+
+  .icon {
+    margin-bottom: 16px;
+    width: 22px;
+    height: 22px;
+  }
+
+  @media (max-width: 1024px) {
+    flex-direction: column;
+    text-align: center;
+  }
+}
+
+#docs {
+  border-right: 1px solid var(--border);
+
+  @media (max-width: 1024px) {
+    border-right: none;
+    border-bottom: 1px solid var(--border);
+  }
+}
+
+#next-steps ul {
+  list-style: none;
+  padding: 0;
+  display: flex;
+  gap: 8px;
+  margin: 32px 0 0;
+
+  .logo {
+    height: 18px;
+  }
+
+  a {
+    color: var(--text-h);
+    font-size: 16px;
+    border-radius: 6px;
+    background: var(--social-bg);
+    display: flex;
+    padding: 6px 12px;
+    align-items: center;
+    gap: 8px;
+    text-decoration: none;
+    transition: box-shadow 0.3s;
+
+    &:hover {
+      box-shadow: var(--shadow);
+    }
+    .button-icon {
+      height: 18px;
+      width: 18px;
+    }
+  }
+
+  @media (max-width: 1024px) {
+    margin-top: 20px;
+    flex-wrap: wrap;
+    justify-content: center;
+
+    li {
+      flex: 1 1 calc(50% - 8px);
+    }
+
+    a {
+      width: 100%;
+      justify-content: center;
+      box-sizing: border-box;
+    }
+  }
+}
+
+#spacer {
+  height: 88px;
+  border-top: 1px solid var(--border);
+  @media (max-width: 1024px) {
+    height: 48px;
+  }
+}
+
+.ticks {
+  position: relative;
+  width: 100%;
+
+  &::before,
+  &::after {
+    content: '';
+    position: absolute;
+    top: -4.5px;
+    border: 5px solid transparent;
+  }
+
+  &::before {
+    left: 0;
+    border-left-color: var(--border);
+  }
+  &::after {
+    right: 0;
+    border-right-color: var(--border);
+  }
+}

frontend/src/App.jsx

@@ -0,0 +1,62 @@
+import { BrowserRouter, Routes, Route, Navigate } from 'react-router-dom'
+import { ConfigProvider } from 'antd'
+import zhCN from 'antd/locale/zh_CN'
+import AppLayout from './components/Layout/AppLayout'
+import Login from './pages/Login/Login'
+import Dashboard from './pages/Dashboard/Dashboard'
+import EditorHome from './pages/EditorHome/EditorHome'
+import TPS from './pages/TPS/TPS'
+import KnowledgeBase from './pages/KnowledgeBase/KnowledgeBase'
+import Doco from './pages/Doco/Doco'
+import UserManage from './pages/UserManage/UserManage'
+import EditorDesk from './pages/EditorDesk/EditorDesk'
+import AuthGuard from './components/AuthGuard/AuthGuard'
+import RoleGuard from './components/AuthGuard/RoleGuard'
+
+const theme = {
+  token: {
+    colorPrimary: '#6b8e6b',
+    borderRadius: 12,
+    borderRadiusLG: 16,
+  },
+}
+
+function App() {
+  return (
+    <ConfigProvider locale={zhCN} theme={theme}>
+      <BrowserRouter>
+        <Routes>
+          <Route path="/login" element={<Login />} />
+          <Route path="/" element={<AuthGuard />}>
+            <Route element={<AppLayout />}>
+              <Route index element={<Navigate to="/dashboard" replace />} />
+              <Route path="dashboard" element={<Dashboard />} />
+              <Route path="tps" element={<TPS />} />
+              <Route path="knowledge" element={<KnowledgeBase />} />
+              <Route path="doco" element={<Doco />} />
+              <Route path="editor-home" element={<EditorHome />} />
+              <Route
+                path="users"
+                element={
+                  <RoleGuard roles={['zhipianren']}>
+                    <UserManage />
+                  </RoleGuard>
+                }
+              />
+              <Route
+                path="editor-desk"
+                element={
+                  <RoleGuard roles={['zhipianren', 'zebian']}>
+                    <EditorDesk />
+                  </RoleGuard>
+                }
+              />
+            </Route>
+          </Route>
+        </Routes>
+      </BrowserRouter>
+    </ConfigProvider>
+  )
+}
+
+export default App

frontend/src/assets/react.svg

@@ -0,0 +1 @@
+<svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" class="iconify iconify--logos" width="35.93" height="32" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 228"><path fill="#00D8FF" d="M210.483 73.824a171.49 171.49 0 0 0-8.24-2.597c.465-1.9.893-3.777 1.273-5.621c6.238-30.281 2.16-54.676-11.769-62.708c-13.355-7.7-35.196.329-57.254 19.526a171.23 171.23 0 0 0-6.375 5.848a155.866 155.866 0 0 0-4.241-3.917C100.759 3.829 77.587-4.822 63.673 3.233C50.33 10.957 46.379 33.89 51.995 62.588a170.974 170.974 0 0 0 1.892 8.48c-3.28.932-6.445 1.924-9.474 2.98C17.309 83.498 0 98.307 0 113.668c0 15.865 18.582 31.778 46.812 41.427a145.52 145.52 0 0 0 6.921 2.165a167.467 167.467 0 0 0-2.01 9.138c-5.354 28.2-1.173 50.591 12.134 58.266c13.744 7.926 36.812-.22 59.273-19.855a145.567 145.567 0 0 0 5.342-4.923a168.064 168.064 0 0 0 6.92 6.314c21.758 18.722 43.246 26.282 56.54 18.586c13.731-7.949 18.194-32.003 12.4-61.268a145.016 145.016 0 0 0-1.535-6.842c1.62-.48 3.21-.974 4.76-1.488c29.348-9.723 48.443-25.443 48.443-41.52c0-15.417-17.868-30.326-45.517-39.844Zm-6.365 70.984c-1.4.463-2.836.91-4.3 1.345c-3.24-10.257-7.612-21.163-12.963-32.432c5.106-11 9.31-21.767 12.459-31.957c2.619.758 5.16 1.557 7.61 2.4c23.69 8.156 38.14 20.213 38.14 29.504c0 9.896-15.606 22.743-40.946 31.14Zm-10.514 20.834c2.562 12.94 2.927 24.64 1.23 33.787c-1.524 8.219-4.59 13.698-8.382 15.893c-8.067 4.67-25.32-1.4-43.927-17.412a156.726 156.726 0 0 1-6.437-5.87c7.214-7.889 14.423-17.06 21.459-27.246c12.376-1.098 24.068-2.894 34.671-5.345a134.17 134.17 0 0 1 1.386 6.193ZM87.276 214.515c-7.882 2.783-14.16 2.863-17.955.675c-8.075-4.657-11.432-22.636-6.853-46.752a156.923 156.923 0 0 1 1.869-8.499c10.486 2.32 22.093 3.988 34.498 4.994c7.084 9.967 14.501 19.128 21.976 27.15a134.668 134.668 0 0 1-4.877 4.492c-9.933 8.682-19.886 14.842-28.658 17.94ZM50.35 144.747c-12.483-4.267-22.792-9.812-29.858-15.863c-6.35-5.437-9.555-10.836-9.555-15.216c0-9.322 13.897-21.212 37.076-29.293c2.813-.98 5.757-1.905 8.812-2.773c3.204 10.42 7.406 21.315 12.477 32.332c-5.137 11.18-9.399 22.249-12.634 32.792a134.718 134.718 0 0 1-6.318-1.979Zm12.378-84.26c-4.811-24.587-1.616-43.134 6.425-47.789c8.564-4.958 27.502 2.111 47.463 19.835a144.318 144.318 0 0 1 3.841 3.545c-7.438 7.987-14.787 17.08-21.808 26.988c-12.04 1.116-23.565 2.908-34.161 5.309a160.342 160.342 0 0 1-1.76-7.887Zm110.427 27.268a347.8 347.8 0 0 0-7.785-12.803c8.168 1.033 15.994 2.404 23.343 4.08c-2.206 7.072-4.956 14.465-8.193 22.045a381.151 381.151 0 0 0-7.365-13.322Zm-45.032-43.861c5.044 5.465 10.096 11.566 15.065 18.186a322.04 322.04 0 0 0-30.257-.006c4.974-6.559 10.069-12.652 15.192-18.18ZM82.802 87.83a323.167 323.167 0 0 0-7.227 13.238c-3.184-7.553-5.909-14.98-8.134-22.152c7.304-1.634 15.093-2.97 23.209-3.984a321.524 321.524 0 0 0-7.848 12.897Zm8.081 65.352c-8.385-.936-16.291-2.203-23.593-3.793c2.26-7.3 5.045-14.885 8.298-22.6a321.187 321.187 0 0 0 7.257 13.246c2.594 4.48 5.28 8.868 8.038 13.147Zm37.542 31.03c-5.184-5.592-10.354-11.779-15.403-18.433c4.902.192 9.899.29 14.978.29c5.218 0 10.376-.117 15.453-.343c-4.985 6.774-10.018 12.97-15.028 18.486Zm52.198-57.817c3.422 7.8 6.306 15.345 8.596 22.52c-7.422 1.694-15.436 3.058-23.88 4.071a382.417 382.417 0 0 0 7.859-13.026a347.403 347.403 0 0 0 7.425-13.565Zm-16.898 8.101a358.557 358.557 0 0 1-12.281 19.815a329.4 329.4 0 0 1-23.444.823c-7.967 0-15.716-.248-23.178-.732a310.202 310.202 0 0 1-12.513-19.846h.001a307.41 307.41 0 0 1-10.923-20.627a310.278 310.278 0 0 1 10.89-20.637l-.001.001a307.318 307.318 0 0 1 12.413-19.761c7.613-.576 15.42-.876 23.31-.876H128c7.926 0 15.743.303 23.354.883a329.357 329.357 0 0 1 12.335 19.695a358.489 358.489 0 0 1 11.036 20.54a329.472 329.472 0 0 1-11 20.722Zm22.56-122.124c8.572 4.944 11.906 24.881 6.52 51.026c-.344 1.668-.73 3.367-1.15 5.09c-10.622-2.452-22.155-4.275-34.23-5.408c-7.034-10.017-14.323-19.124-21.64-27.008a160.789 160.789 0 0 1 5.888-5.4c18.9-16.447 36.564-22.941 44.612-18.3ZM128 90.808c12.625 0 22.86 10.235 22.86 22.86s-10.235 22.86-22.86 22.86s-22.86-10.235-22.86-22.86s10.235-22.86 22.86-22.86Z"></path></svg>

frontend/src/components/AuthGuard/AuthGuard.jsx

@@ -0,0 +1,21 @@
+import { useEffect } from 'react'
+import { Navigate, Outlet } from 'react-router-dom'
+import useAuthStore from '../../stores/authStore'
+
+function AuthGuard() {
+  const { user, fetchMe } = useAuthStore()
+
+  useEffect(() => {
+    if (user === null) {
+      fetchMe()
+    }
+  }, [user, fetchMe])
+
+  if (user === null) {
+    return null
+  }
+
+  return user ? <Outlet /> : <Navigate to="/login" replace />
+}
+
+export default AuthGuard

frontend/src/components/AuthGuard/RoleGuard.jsx

@@ -0,0 +1,20 @@
+/**
+ * 角色路由守卫 — 拦截非授权角色访问受保护路由
+ * 用法: <Route path="/users" element={<RoleGuard roles={['zhipianren']}><UserManage /></RoleGuard>} />
+ */
+import { Navigate } from 'react-router-dom'
+import useAuthStore from '../../stores/authStore'
+
+function RoleGuard({ children, roles }) {
+  const { user } = useAuthStore()
+
+  if (!user) return null // AuthGuard 已处理未登录跳转
+
+  if (!roles.includes(user.role)) {
+    return <Navigate to="/dashboard" replace />
+  }
+
+  return children
+}
+
+export default RoleGuard

frontend/src/components/KnowledgeTree/KnowledgeTree.css

@@ -0,0 +1,44 @@
+.kt-container {
+  display: flex;
+  flex-direction: column;
+  height: 100%;
+}
+
+.kt-header {
+  padding: 8px 12px;
+  border-bottom: 1px solid #f0f0f0;
+  background: #fafafa;
+}
+
+.kt-tree-wrapper {
+  flex: 1;
+  overflow-y: auto;
+  padding: 8px 4px;
+}
+
+.kt-empty {
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  height: 120px;
+  color: #999;
+  font-size: 13px;
+}
+
+/* 节点文字 */
+.kt-node-title {
+  font-size: 13px;
+  color: #3b4a3b;
+}
+
+/* 全部根节点高亮 */
+.kt-node-root {
+  font-weight: 600;
+  color: #3b4a3b;
+}
+
+/* Tree 选中态 */
+.kt-container .ant-tree-node-selected .kt-node-title {
+  color: #6b8e6b;
+  font-weight: 600;
+}

frontend/src/components/KnowledgeTree/KnowledgeTree.jsx

@@ -0,0 +1,147 @@
+/**
+ * 知识库树形导航组件（交互层）
+ *
+ * 职责：
+ * - 树的展开/收起/记忆状态
+ * - 节点高亮 + 联动回调
+ * - 「全部展开 / 全部收起」操作按钮
+ *
+ * 数据分组逻辑完全委托给后端 get_grouped_items（数据分组层），
+ * 本组件只管交互，不感知分组的业务含义。
+ *
+ * 二级节点 key 格式（由后端控制）：
+ * - 大类节点：                    "manuscript"
+ * - 二级节点（新格式）：           "manuscript|author|左鑫"
+ *                                 type|二级字段名|字段值
+ * - 二级节点（出处，杂志文章）：  "military_report|source_detail|航空知识..."
+ *
+ * 回调给 KnowledgeBase：统一返回 { type, detail }
+ * - 大类节点 → { type: "manuscript", detail: null }
+ * - 作者二级节点 → { type: "manuscript", detail: "左鑫" }
+ * - 出处二级节点 → { type: "military_report", detail: "航空知识..." }
+ * （detail=author名 或 source_detail原文，与 displayedItems 过滤逻辑对齐）
+ */
+
+import { useState, useCallback } from 'react'
+import { Tree, Button, Space } from 'antd'
+import { CaretDownFilled, NodeIndexOutlined } from '@ant-design/icons'
+import './KnowledgeTree.css'
+
+const { TreeNode } = Tree
+
+export default function KnowledgeTree({
+  treeData,         // 来自 getGroupedItems() API 的原始树数据（含全部节点）
+  onNodeSelect,      // 选中节点时回调，参数: { key, type, detail } | null（全部）
+  selectedKey,       // 当前选中的 key（用于高亮）
+}) {
+  const [expandedKeys, setExpandedKeys] = useState(() => {
+    // 默认全部展开
+    if (!treeData || treeData.length === 0) return []
+    const root = treeData[0]
+    return [root.key, ...(root.children || []).map(c => c.key)]
+  })
+
+  // 全部展开
+  const handleExpandAll = useCallback(() => {
+    if (!treeData || treeData.length === 0) return
+    const root = treeData[0]
+    const allKeys = [root.key]
+    for (const child of root.children || []) {
+      allKeys.push(child.key)
+      if (child.children) {
+        for (const grandchild of child.children) {
+          allKeys.push(grandchild.key)
+        }
+      }
+    }
+    setExpandedKeys(allKeys)
+  }, [treeData])
+
+  // 全部收起
+  const handleCollapseAll = useCallback(() => {
+    setExpandedKeys([])
+  }, [])
+
+  // 节点选择
+  const handleSelect = useCallback((selectedKeys, info) => {
+    if (selectedKeys.length === 0) {
+      onNodeSelect(null)
+      return
+    }
+    const key = selectedKeys[0]
+
+    // 解析 key 格式：
+    // "all"                                → 全部根节点
+    // "manuscript"                         → 大类节点
+    // "manuscript|author|左鑫"             → 二级节点（新格式：type|fieldName|fieldValue）
+    // "military_report|source_detail|..."  → 二级节点（出处）
+    if (key === 'all') {
+      onNodeSelect(null)
+      return
+    }
+
+    const parts = key.split('|')
+    if (parts.length === 1) {
+      // 大类节点
+      onNodeSelect({ key, type: parts[0], detail: null })
+    } else {
+      // 二级节点：parts[0]=type, parts[1]=字段名(废弃不用), parts[2]=字段值
+      // 统一以 parts[2] 作为 detail（作者名 或 出处名），与 displayedItems 过滤逻辑对齐
+      onNodeSelect({ key, type: parts[0], detail: parts[2] })
+    }
+  }, [onNodeSelect])
+
+  // 渲染节点
+  const renderNode = (node, isRoot = false) => {
+    return (
+      <TreeNode
+        key={node.key}
+        title={
+          <span className={`kt-node-title ${isRoot ? 'kt-node-root' : ''}`}>
+            {node.label}
+          </span>
+        }
+        icon={isRoot ? <NodeIndexOutlined /> : undefined}
+      >
+        {(node.children || []).map(child => renderNode(child, false))}
+      </TreeNode>
+    )
+  }
+
+  if (!treeData || treeData.length === 0) {
+    return (
+      <div className="kt-empty">
+        暂无知识库条目
+      </div>
+    )
+  }
+
+  return (
+    <div className="kt-container">
+      <div className="kt-header">
+        <Space size="small">
+          <Button size="small" onClick={handleExpandAll}>
+            全部展开
+          </Button>
+          <Button size="small" onClick={handleCollapseAll}>
+            全部收起
+          </Button>
+        </Space>
+      </div>
+      <div className="kt-tree-wrapper">
+        <Tree
+          showIcon
+          showLine={{ showLeafIcon: false }}
+          expandedKeys={expandedKeys}
+          selectedKeys={selectedKey ? [selectedKey] : []}
+          onExpand={(keys) => setExpandedKeys(keys)}
+          onSelect={handleSelect}
+          blockNode
+          switcherIcon={<CaretDownFilled />}
+        >
+          {treeData.map(node => renderNode(node, true))}
+        </Tree>
+      </div>
+    </div>
+  )
+}

frontend/src/components/Layout/AppLayout.css

@@ -0,0 +1,34 @@
+.app-layout {
+  min-height: 100vh;
+}
+
+.app-sider {
+  background: #fff !important;
+  border-right: 1px solid #f0f0f0;
+  width: 220px;
+  flex: 0 0 220px;
+  overflow-y: auto;
+}
+
+.app-main {
+  min-height: 100vh;
+}
+
+.app-header {
+  background: #fff !important;
+  border-bottom: 1px solid #f0f0f0;
+  padding: 0 24px !important;
+  height: 64px;
+  line-height: 64px;
+  position: sticky;
+  top: 0;
+  z-index: 100;
+}
+
+.app-content {
+  padding: 24px;
+  background: var(--color-bg-cream);
+  min-height: calc(100vh - 64px);
+  max-width: 1400px;
+  margin: 0 auto;
+}

frontend/src/components/Layout/AppLayout.jsx

@@ -0,0 +1,27 @@
+import { Layout } from 'antd'
+import { Outlet } from 'react-router-dom'
+import SideNav from './SideNav'
+import TopBar from './TopBar'
+import './AppLayout.css'
+
+const { Sider, Header, Content } = Layout
+
+function AppLayout() {
+  return (
+    <Layout className="app-layout">
+      <Sider width={220} className="app-sider">
+        <SideNav />
+      </Sider>
+      <Layout className="app-main">
+        <Header className="app-header">
+          <TopBar />
+        </Header>
+        <Content className="app-content">
+          <Outlet />
+        </Content>
+      </Layout>
+    </Layout>
+  )
+}
+
+export default AppLayout

frontend/src/components/Layout/SideNav.css

@@ -0,0 +1,62 @@
+.side-nav {
+  padding: 16px 0;
+  height: 100%;
+}
+
+.side-nav-logo {
+  padding: 16px 24px 24px;
+  border-bottom: 1px solid #f0f0f0;
+  margin-bottom: 8px;
+}
+
+.logo-text {
+  display: block;
+  font-size: 18px;
+  font-weight: 600;
+  color: var(--color-primary-green);
+}
+
+.logo-sub {
+  display: block;
+  font-size: 12px;
+  color: var(--color-text-secondary);
+  margin-top: 2px;
+}
+
+.side-nav-menu {
+  border-right: none !important;
+}
+
+.side-nav-menu .ant-menu-item {
+  margin: 4px 8px;
+  border-radius: 8px;
+}
+
+.side-nav-menu .ant-menu-item-selected {
+  background: var(--color-accent-green) !important;
+  color: var(--color-primary-green) !important;
+}
+
+/* disabled 项整体灰化 + 不可点击 */
+.side-nav-menu .ant-menu-item-disabled {
+  opacity: 0.55;
+  cursor: not-allowed !important;
+}
+
+/* 即将上线小标签 */
+.menu-soon-tag {
+  font-size: 10px;
+  color: #999;
+  background: #f0f0f0;
+  padding: 1px 5px;
+  border-radius: 4px;
+  margin-left: 6px;
+  vertical-align: middle;
+  white-space: nowrap;
+}
+
+/* disabled 项内标签不受 opacity 影响 */
+.side-nav-menu .ant-menu-item-disabled .menu-soon-tag {
+  opacity: 1;
+  color: #888;
+}

frontend/src/components/Layout/SideNav.jsx

@@ -0,0 +1,84 @@
+import { Menu } from 'antd'
+import { useNavigate, useLocation } from 'react-router-dom'
+import {
+  DashboardOutlined,
+  FileTextOutlined,
+  BookOutlined,
+  SyncOutlined,
+  UserOutlined,
+  TeamOutlined,
+  SoundOutlined,
+} from '@ant-design/icons'
+import useAuthStore from '../../stores/authStore'
+
+function SideNav() {
+  const navigate = useNavigate()
+  const location = useLocation()
+  const { user } = useAuthStore()
+
+  const allMenuItems = [
+    { key: '/dashboard', icon: <DashboardOutlined />, label: '仪表盘' },
+    { key: '/editor-desk', icon: <FileTextOutlined />, label: '责编录入' },
+    { key: '/tps', icon: <FileTextOutlined />, label: 'TPS 选题策划' },
+    { key: '/knowledge', icon: <BookOutlined />, label: '知识库' },
+    { key: '/doco', icon: <SyncOutlined />, label: '文稿对齐' },
+    { key: '/editor-home', icon: <UserOutlined />, label: '个人首页' },
+    { key: '/users', icon: <TeamOutlined />, label: '用户管理' },
+    {
+      key: 'tts-placeholder',
+      icon: <SoundOutlined />,
+      label: (
+        <span className="menu-item-with-tag">
+          蓝皓配音 TTS 2.0
+          <span className="menu-soon-tag">即将上线</span>
+        </span>
+      ),
+      disabled: true,
+    },
+    {
+      key: 'collab-placeholder',
+      icon: <TeamOutlined />,
+      label: (
+        <span className="menu-item-with-tag">
+          内部协作（Mattermost）
+          <span className="menu-soon-tag">即将上线</span>
+        </span>
+      ),
+      disabled: true,
+    },
+  ]
+
+  // 按角色过滤菜单项
+  // user.role 值来自后端 /api/auth/me，应为字符串 'zhipianren' / 'zebian' / 'biandao'
+  const visibleItems = allMenuItems.filter(item => {
+    // /dashboard 三角色都可见
+    if (item.key === '/dashboard') return true
+    // /users 仅 zhipianren
+    if (item.key === '/users') return String(user?.role) === 'zhipianren'
+    // /editor-desk 仅 zhipianren + zebian
+    if (item.key === '/editor-desk') {
+      const role = String(user?.role)
+      return role === 'zhipianren' || role === 'zebian'
+    }
+    // 其余三角色都可见（保留后期 Phase 入口）
+    return true
+  })
+
+  return (
+    <div className="side-nav">
+      <div className="side-nav-logo">
+        <span className="logo-text">军事科技</span>
+        <span className="logo-sub">TPS 工作台</span>
+      </div>
+      <Menu
+        mode="inline"
+        selectedKeys={[location.pathname]}
+        items={visibleItems}
+        onClick={({ key }) => navigate(key)}
+        className="side-nav-menu"
+      />
+    </div>
+  )
+}
+
+export default SideNav

frontend/src/components/Layout/TopBar.css

@@ -0,0 +1,27 @@
+.top-bar {
+  display: flex;
+  align-items: center;
+  justify-content: space-between;
+  height: 100%;
+}
+
+.top-bar-search {
+  flex: 1;
+  max-width: 400px;
+}
+
+.top-bar-actions {
+  display: flex;
+  align-items: center;
+  gap: 16px;
+}
+
+.top-bar-icon {
+  font-size: 18px;
+  color: var(--color-text-secondary);
+  cursor: pointer;
+}
+
+.top-bar-avatar {
+  cursor: pointer;
+}

frontend/src/components/Layout/TopBar.jsx

@@ -0,0 +1,38 @@
+import { Input, Dropdown, Avatar } from 'antd'
+import { SearchOutlined, BellOutlined, LogoutOutlined } from '@ant-design/icons'
+import useAuthStore from '../../stores/authStore'
+import './TopBar.css'
+
+function TopBar() {
+  const { user, logout } = useAuthStore()
+
+  const items = [
+    {
+      key: 'logout',
+      icon: <LogoutOutlined />,
+      label: '退出',
+      onClick: () => logout(),
+    },
+  ]
+
+  const displayName = user?.display_name || user?.username || '未知用户'
+  const initials = displayName.slice(0, 1).toUpperCase()
+
+  return (
+    <div className="top-bar">
+      <div className="top-bar-search">
+        <Input placeholder="搜索..." prefix={<SearchOutlined />} />
+      </div>
+      <div className="top-bar-actions">
+        <BellOutlined className="top-bar-icon" />
+        <Dropdown menu={{ items }} placement="bottomRight">
+          <Avatar className="top-bar-avatar" style={{ backgroundColor: 'var(--color-primary-green)' }}>
+            {initials}
+          </Avatar>
+        </Dropdown>
+      </div>
+    </div>
+  )
+}
+
+export default TopBar

frontend/src/hooks/useKnowledgeGrouping.js

@@ -0,0 +1,171 @@
+/**
+ * 知识库数据分组层（交互层与分组维度完全解耦）
+ *
+ * 本模块职责：
+ * - 接收 items 数组，按指定维度生成分组树数据
+ * - 完全不涉及 UI（无 React 组件、无 Ant Design 引用）
+ *
+ * 将来切换为「按主题/装备分组」时：
+ * - 只改本文件中的 groupItemsBySource 函数
+ * - 前端 KnowledgeTree 组件一行都不用动
+ */
+
+/**
+ * ============================================================
+ * 📌 二级分组维度映射表（核心配置，一处可改）
+ * ============================================================
+ * key = source_type，value = 用来做二级分组的字段名（取 items 里的字段）
+ * null = 该大类不建二级节点，条目直接挂在大类下
+ *
+ * 当前配置：
+ * - manuscript（节目文稿）    → 按 author（作者/编导）归堆
+ * - military_report（杂志文章）→ 按 source_detail（出处）归堆
+ * - baoti（报题单）          → 不分组
+ * - manual（其他）            → 不分组
+ *
+ * 将来调整某类的二级维度，只需改这一张表，本文件分组逻辑全部收敛。
+ */
+const SECONDARY_GROUP_FIELD = {
+  manuscript: 'author',           // 节目文稿 → 按作者(编导)
+  military_report: 'source_detail', // 杂志文章 → 按出处
+  baoti: null,                    // 报题单 → 不分组
+  manual: null,                   // 其他 → 不分组
+}
+
+/**
+ * ============================================================
+ * 📌 来源大类固定显示顺序（制片人 Obsidian 习惯）
+ * ============================================================
+ * 即使某类暂时为空也按此顺序； 有数据时即按此次序排列。
+ * 已有类别按此次序排，未知类别兜底排末尾。
+ */
+const SOURCE_TYPE_ORDER = [
+  'manuscript',      // 节目文稿
+  'military_report', // 杂志文章
+  'baoti',           // 报题单
+  'manual',          // 其他
+]
+
+/**
+ * source_type 中文标签
+ */
+const SOURCE_TYPE_LABEL = {
+  military_report: '杂志文章',
+  manuscript: '节目文稿',
+  baoti: '报题单',
+  manual: '其他',
+}
+
+/**
+ * 按来源（source_type → 二级字段）分组
+ * @param {Array} items - 知识库条目数组，含 source_type 字段，及 SECONDARY_GROUP_FIELD 中指定的二级字段
+ * @returns {Array} 树形结构数据，供 KnowledgeTree 组件渲染
+ *
+ * 返回格式：
+ * [
+ *   {
+ *     key: "all",
+ *     label: "全部（N条）",
+ *     count: 总条目数,
+ *     children: [
+ *       {
+ *         key: "manuscript",
+ *         label: "节目文稿（n条）",
+ *         count: n,
+ *         children: [  // SECONDARY_GROUP_FIELD[st] 为 null 时为空数组
+ *           { key: "manuscript|author|左鑫", label: "左鑫（3条）", count: 3 }
+ *         ]
+ *       },
+ *       ...
+ *     ]
+ *   }
+ * ]
+ *
+ * 二级节点 key 格式：`{source_type}|{二级字段名}|{字段值}`
+ * 例：manuscript|author|左鑫
+ *     military_report|source_detail|航空知识 2026年第1期
+ *
+ * 注意：
+ * - 二级字段值为 null / 空字串 → 归入对应大类，不造空节点
+ * - 无效的 source_type → 归入 'manual' 大类
+ */
+export function groupItemsBySource(items) {
+  const totalCount = items.length
+
+  // 按 source_type 分组（按固定顺序遍历）
+  const typeGroups = {}
+  // 初始化所有已知类别，确保空类别也出现在树中
+  for (const st of SOURCE_TYPE_ORDER) {
+    typeGroups[st] = []
+  }
+
+  for (const item of items) {
+    const st = item.source_type || 'manual'
+    // 未知类别兜底（如 'journal' 等新加类型）
+    if (!typeGroups[st]) {
+      typeGroups[st] = []
+    }
+    typeGroups[st].push(item)
+  }
+
+  const children = []
+
+  // 按固定顺序遍历 SOURCE_TYPE_ORDER
+  for (const st of SOURCE_TYPE_ORDER) {
+    const stItems = typeGroups[st] || []
+    if (stItems.length === 0) {
+      // 空类别也按顺序出现在树中（0条）
+      children.push({
+        key: st,
+        label: `${SOURCE_TYPE_LABEL[st] || st}（0条）`,
+        count: 0,
+        children: [],
+      })
+      continue
+    }
+
+    const secondaryField = SECONDARY_GROUP_FIELD[st] || null
+    const grandchildren = []
+
+    if (secondaryField !== null) {
+      // 按二级字段分组
+      const detailGroups = {}
+      for (const item of stItems) {
+        // 取二级字段值，null/空字串统一为 null 处理（不造空节点）
+        const sd = (item[secondaryField] || '').trim() || null
+        if (!detailGroups[sd]) {
+          detailGroups[sd] = []
+        }
+        detailGroups[sd].push(item)
+      }
+
+      for (const [sd, sdItems] of Object.entries(detailGroups)) {
+        if (sd !== null) {
+          // 二级节点 key 格式：type|fieldName|fieldValue
+          grandchildren.push({
+            key: `${st}|${secondaryField}|${sd}`,
+            label: `${sd}（${sdItems.length}条）`,
+            count: sdItems.length,
+          })
+        }
+      }
+    }
+    // secondaryField 为 null 时，grandchildren 保持为空数组 → 大类直接可点击
+
+    children.push({
+      key: st,
+      label: `${SOURCE_TYPE_LABEL[st] || st}（${stItems.length}条）`,
+      count: stItems.length,
+      children: grandchildren,
+    })
+  }
+
+  return [
+    {
+      key: 'all',
+      label: `全部（${totalCount}条）`,
+      count: totalCount,
+      children,
+    },
+  ]
+}

frontend/src/index.css

@@ -0,0 +1,34 @@
+:root {
+  /* 奶油底，主背景 */
+  --color-bg-cream: #fbf9f1;
+  /* 深绿，主色 */
+  --color-primary-green: #6b8e6b;
+  /* 浅绿，辅色 */
+  --color-accent-green: #c8d9b0;
+  /* 文字主色 */
+  --color-text-primary: #1a1a1a;
+  /* 文字次要色 */
+  --color-text-secondary: #6b6b6b;
+  /* 卡片大圆角 */
+  --radius-card: 16px;
+  /* 按钮圆角 */
+  --radius-button: 12px;
+}
+
+* {
+  box-sizing: border-box;
+}
+
+body {
+  margin: 0;
+  background-color: var(--color-bg-cream);
+  font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', Roboto, 'Helvetica Neue', Arial, sans-serif;
+  color: var(--color-text-primary);
+  -webkit-font-smoothing: antialiased;
+  -moz-osx-font-smoothing: grayscale;
+}
+
+#root {
+  min-height: 100vh;
+  width: 100%;
+}

frontend/src/main.jsx

@@ -0,0 +1,10 @@
+import { StrictMode } from 'react'
+import { createRoot } from 'react-dom/client'
+import './index.css'
+import App from './App.jsx'
+
+createRoot(document.getElementById('root')).render(
+  <StrictMode>
+    <App />
+  </StrictMode>,
+)

frontend/src/pages/Dashboard/ChangeCoverModal.jsx

@@ -0,0 +1,151 @@
+import { useState, useEffect } from 'react'
+import { Modal, Upload, Select, Button, message } from 'antd'
+import { UploadOutlined, InboxOutlined } from '@ant-design/icons'
+import { uploadCover } from '../../services/dashboardService'
+
+const { Dragger } = Upload
+
+/**
+ * 更换题图 Modal
+ * 
+ * 权限：仅 zhipianren / zebian 可用（由父组件控制显示/隐藏）
+ * 
+ * @param {boolean} open - Modal 是否显示
+ * @param {Function} onClose - 关闭回调
+ * @param {Function} onUploaded - 上传成功回调，传入新题图对象
+ * @param {Array} episodes - 期次列表（用于下拉选择关联期次）
+ */
+function ChangeCoverModal({ open, onClose, onUploaded, episodes = [] }) {
+  const [file, setFile] = useState(null)
+  const [previewUrl, setPreviewUrl] = useState(null)
+  const [selectedEpisode, setSelectedEpisode] = useState(null)
+  const [uploading, setUploading] = useState(false)
+
+  // 每次打开时重置状态
+  useEffect(() => {
+    if (open) {
+      setFile(null)
+      setPreviewUrl(null)
+      setSelectedEpisode(null)
+      setUploading(false)
+    }
+  }, [open])
+
+  // 选择文件后生成预览
+  const handleFileChange = (info) => {
+    const rawFile = info.file.originFileObj || info.file
+    if (!rawFile) return
+    setFile(rawFile)
+    const url = URL.createObjectURL(rawFile)
+    setPreviewUrl(url)
+  }
+
+  const handleUpload = async () => {
+    if (!file) {
+      message.warning('请先选择图片')
+      return
+    }
+    if (!selectedEpisode) {
+      message.warning('请选择关联期次')
+      return
+    }
+
+    setUploading(true)
+    try {
+      const ep = episodes.find(e => String(e.episode_number) === String(selectedEpisode))
+      const result = await uploadCover(file, Number(selectedEpisode), ep?.program_name || `第${selectedEpisode}期`)
+      message.success('题图上传成功')
+      onUploaded({
+        cover_path: result.cover_path,
+        episode_number: Number(selectedEpisode),
+        episode_title: ep?.program_name || `第${selectedEpisode}期`,
+      })
+    } catch (err) {
+      const errMsg = err?.response?.data?.detail || '上传失败，请重试'
+      message.error(errMsg)
+    } finally {
+      setUploading(false)
+    }
+  }
+
+  const episodeOptions = episodes.map(ep => ({
+    value: String(ep.episode_number),
+    label: `第 ${ep.episode_number} 期 · ${ep.program_name}`,
+  }))
+
+  return (
+    <Modal
+      title="更换题图"
+      open={open}
+      onCancel={onClose}
+      footer={null}
+      width={480}
+      destroyOnClose
+    >
+      <div style={{ display: 'flex', flexDirection: 'column', gap: 16 }}>
+        {/* 图片上传区域 */}
+        <div>
+          <p style={{ fontSize: 13, color: '#6b6b6b', marginBottom: 8 }}>上传海报图片（PNG/JPG/WEBP，不超过 5MB）</p>
+          <Dragger
+            accept=".png,.jpg,.jpeg,.webp"
+            maxCount={1}
+            beforeUpload={() => false}
+            onChange={handleFileChange}
+            showUploadList={false}
+            style={{ borderRadius: 12 }}
+          >
+            {previewUrl ? (
+              <div style={{ padding: '8px 0' }}>
+                <img
+                  src={previewUrl}
+                  alt="预览"
+                  style={{ maxWidth: '100%', maxHeight: 160, borderRadius: 8 }}
+                />
+                <p style={{ marginTop: 8, color: '#6b8e6b', fontSize: 13 }}>点击或拖拽更换图片</p>
+              </div>
+            ) : (
+              <>
+                <p className="ant-upload-drag-icon" style={{ marginBottom: 8 }}>
+                  <InboxOutlined style={{ fontSize: 32, color: '#6b8e6b' }} />
+                </p>
+                <p style={{ color: '#6b6b6b', fontSize: 13 }}>点击或拖拽图片到此处上传</p>
+              </>
+            )}
+          </Dragger>
+        </div>
+
+        {/* 关联期次下拉 */}
+        <div>
+          <p style={{ fontSize: 13, color: '#6b6b6b', marginBottom: 8 }}>关联期次</p>
+          <Select
+            placeholder="请选择期次"
+            style={{ width: '100%' }}
+            value={selectedEpisode}
+            onChange={setSelectedEpisode}
+            options={episodeOptions}
+            showSearch
+            filterOption={(input, option) =>
+              option.label.toLowerCase().includes(input.toLowerCase())
+            }
+          />
+        </div>
+
+        {/* 操作按钮 */}
+        <div style={{ display: 'flex', gap: 8, justifyContent: 'flex-end' }}>
+          <Button onClick={onClose} disabled={uploading}>取消</Button>
+          <Button
+            type="primary"
+            icon={<UploadOutlined />}
+            loading={uploading}
+            onClick={handleUpload}
+            disabled={!file || !selectedEpisode}
+          >
+            确认上传
+          </Button>
+        </div>
+      </div>
+    </Modal>
+  )
+}
+
+export default ChangeCoverModal

frontend/src/pages/Dashboard/Dashboard.css

@@ -0,0 +1,355 @@
+.dashboard {
+  max-width: 1200px;
+  padding: 12px;
+}
+
+/* ===== Banner ===== */
+.dashboard-banner {
+  width: 100%;
+  height: 150px;
+  border-radius: 14px;
+  overflow: hidden;
+  position: relative;
+  background: #3a4a3a;
+  margin-bottom: 12px;
+}
+
+.banner-bg {
+  position: absolute;
+  inset: 0;
+  object-fit: cover;
+  width: 100%;
+  height: 100%;
+}
+
+.banner-gradient {
+  position: absolute;
+  inset: 0;
+  background: linear-gradient(
+    to right,
+    rgba(12, 16, 12, 0.86) 0%,
+    rgba(12, 16, 12, 0.6) 35%,
+    rgba(12, 16, 12, 0.08) 70%,
+    transparent 100%
+  );
+}
+
+.banner-logo {
+  position: absolute;
+  top: 16px;
+  right: 16px;
+  height: 64px;
+  width: auto;
+  opacity: 0.9;
+}
+
+.banner-text {
+  position: absolute;
+  bottom: 20px;
+  left: 20px;
+  z-index: 2;
+}
+
+.banner-eyebrow {
+  font-size: 11px;
+  color: #cdd8c8;
+  letter-spacing: 1px;
+  margin-bottom: 4px;
+}
+
+.banner-title {
+  font-size: 21px;
+  font-weight: 500;
+  color: #ffffff;
+  margin: 0 0 4px;
+  line-height: 1.2;
+}
+
+.banner-subtitle {
+  font-size: 12px;
+  color: #d6e0d0;
+  margin: 0 0 4px;
+}
+
+.banner-hint {
+  font-size: 11px;
+  color: #bcc8b6;
+  margin: 0;
+}
+
+.change-cover-btn {
+  position: absolute;
+  bottom: 16px;
+  right: 16px;
+  z-index: 3;
+}
+
+/* 默认渐变底图（无题图时） */
+.banner-default-bg {
+  position: absolute;
+  inset: 0;
+  background: linear-gradient(135deg, #3a4a3a 0%, #5a7a5a 50%, #7a9a6a 100%);
+}
+
+/* ===== KPI Strip ===== */
+.kpi-strip {
+  display: flex;
+  gap: 12px;
+  margin-bottom: 12px;
+}
+
+.kpi-chip {
+  flex: 1;
+  display: flex;
+  align-items: center;
+  gap: 10px;
+  background: #fff;
+  border-radius: 12px;
+  padding: 10px 16px;
+  height: 48px;
+  box-shadow: 0 1px 4px rgba(0, 0, 0, 0.06);
+}
+
+.kpi-icon {
+  width: 28px;
+  height: 28px;
+  border-radius: 8px;
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  flex-shrink: 0;
+}
+
+.kpi-label {
+  font-size: 11px;
+  color: #6b6b6b;
+  margin: 0;
+  white-space: nowrap;
+}
+
+.kpi-value {
+  font-size: 15px;
+  font-weight: 600;
+  color: #1a1a1a;
+  margin: 0;
+}
+
+/* ===== Bar Chart Card ===== */
+.bar-chart-card {
+  background: #fff;
+  border-radius: 14px;
+  padding: 16px;
+  margin-bottom: 12px;
+}
+
+.bar-chart-header {
+  display: flex;
+  justify-content: space-between;
+  align-items: center;
+  margin-bottom: 12px;
+}
+
+.bar-chart-title {
+  font-size: 15px;
+  font-weight: 500;
+  color: #3b4a3b;
+  margin: 0;
+}
+
+.bar-chart-hint {
+  font-size: 11px;
+  color: #888780;
+}
+
+.bars-wrapper {
+  display: flex;
+  justify-content: space-between;
+  align-items: flex-end;
+  height: 180px;
+  padding: 0 4px;
+  gap: 4px;
+}
+
+.bar-item {
+  display: flex;
+  flex-direction: column;
+  align-items: center;
+  flex: 1;
+  max-width: 42px;
+  min-width: 20px;
+  justify-content: flex-end;
+  height: 180px;
+}
+
+.bar {
+  width: 100%;
+  max-width: 38px;
+  border-radius: 7px 7px 0 0;
+  transition: all 0.3s;
+  cursor: pointer;
+  position: relative;
+}
+
+.bar-number {
+  font-size: 9px;
+  font-weight: 600;
+  text-align: center;
+  white-space: nowrap;
+  margin-bottom: 2px;
+}
+
+.bar-program {
+  font-size: 10px;
+  color: #6b6b6b;
+  text-align: center;
+  max-width: 64px;
+  overflow: hidden;
+  white-space: nowrap;
+  text-overflow: ellipsis;
+  margin-top: 4px;
+}
+
+.chart-legend {
+  display: flex;
+  justify-content: center;
+  gap: 16px;
+  margin-top: 12px;
+  padding-top: 12px;
+  border-top: 1px solid #f0f0f0;
+}
+
+.legend-item {
+  display: flex;
+  align-items: center;
+  gap: 4px;
+  font-size: 11px;
+  color: #6b6b6b;
+}
+
+.dot {
+  width: 8px;
+  height: 8px;
+  border-radius: 50%;
+}
+
+/* ===== Bottom Two Columns ===== */
+.bottom-row {
+  display: grid;
+  grid-template-columns: 1fr 1fr;
+  gap: 12px;
+}
+
+.side-card {
+  background: #fff;
+  border-radius: 14px;
+  padding: 16px;
+}
+
+.side-card-title {
+  font-size: 15px;
+  font-weight: 500;
+  color: #3b4a3b;
+  margin: 0 0 12px;
+}
+
+/* Radar placeholder card */
+.radar-placeholder {
+  border: 1.5px dashed #c8d9b0;
+  border-radius: 12px;
+  display: flex;
+  flex-direction: column;
+  align-items: center;
+  justify-content: center;
+  min-height: 120px;
+  padding: 16px;
+  position: relative;
+}
+
+.radar-placeholder-label {
+  position: absolute;
+  top: 8px;
+  right: 8px;
+  font-size: 10px;
+  color: #bcc8b6;
+  background: #fbf9f1;
+  padding: 2px 6px;
+  border-radius: 6px;
+}
+
+.radar-placeholder p {
+  font-size: 13px;
+  color: #6b6b6b;
+  margin: 8px 0 4px;
+}
+
+.radar-placeholder small {
+  font-size: 11px;
+  color: #999;
+}
+
+.radar-sample-items {
+  width: 100%;
+  margin-bottom: 8px;
+}
+
+.radar-sample-item {
+  font-size: 12px;
+  color: #4a5a4a;
+  padding: 4px 8px;
+  background: #f0f7ec;
+  border-radius: 6px;
+  margin-bottom: 4px;
+}
+
+/* Schedule list */
+.schedule-list {
+  list-style: none;
+  padding: 0;
+  margin: 0;
+}
+
+.schedule-item {
+  display: flex;
+  justify-content: space-between;
+  align-items: center;
+  padding: 8px 0;
+  border-bottom: 1px solid #f5f5f5;
+  font-size: 12px;
+}
+
+.schedule-item:last-child {
+  border-bottom: none;
+}
+
+.schedule-ep {
+  color: #1a1a1a;
+  font-weight: 500;
+}
+
+.schedule-name {
+  color: #4a5a4a;
+  flex: 1;
+  margin: 0 8px;
+  overflow: hidden;
+  white-space: nowrap;
+  text-overflow: ellipsis;
+}
+
+.schedule-date {
+  color: #888780;
+  font-size: 11px;
+}
+
+.schedule-editor {
+  color: #6b6b6b;
+  font-size: 11px;
+}
+
+/* Loading state */
+.chart-loading {
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  height: 180px;
+  color: #999;
+}

frontend/src/pages/Dashboard/Dashboard.jsx

@@ -0,0 +1,343 @@
+import { useState, useEffect } from 'react'
+import { Row, Col, Card, Avatar, Tooltip, Button, Spin } from 'antd'
+import {
+  EyeOutlined,
+  FireOutlined,
+  CalendarOutlined,
+  BarChartOutlined,
+  PictureOutlined,
+  UploadOutlined,
+  LineChartOutlined,
+  AimOutlined,
+} from '@ant-design/icons'
+import useAuthStore from '../../stores/authStore'
+import { listEpisodes } from '../../services/episodeService'
+import { listTargets } from '../../services/yearlyTargetService'
+import { getCurrentCover, getUpcomingSchedules } from '../../services/dashboardService'
+import ChangeCoverModal from './ChangeCoverModal'
+import './Dashboard.css'
+
+/**
+ * 收视份额颜色判定（极易写反，禁止凭直觉）：
+ *   audience_share > stretch_target → 🔴 红（优秀）
+ *   base_target ≤ audience_share ≤ stretch_target → 🔵 蓝（达标）
+ *   audience_share < base_target → 🟢 绿（待提升）
+ * 判色取该期 air_date 所属年份的 yearly_targets 行（不是当前年）。
+ */
+function getShareColor(share, targets, airYear) {
+  const n = Number(share)
+  const target = (targets || []).find(t => Number(t.year) === Number(airYear))
+  if (!target) return '#999'
+  if (isNaN(n)) return '#999'
+  const stretch = Number(target.stretch_target)
+  const base = Number(target.base_target)
+  if (n > stretch) return '#c0584f'   // 红=超过摸高目标（优秀）
+  if (n >= base) return '#5b8db8'     // 蓝=未达摸高目标（达标）
+  return '#7aa874'                    // 绿=未达基础目标（待提升）
+}
+
+function getColorLabel(share, targets, airYear) {
+  const n = Number(share)
+  const target = (targets || []).find(t => Number(t.year) === Number(airYear))
+  if (!target) return '未知'
+  const stretch = Number(target.stretch_target)
+  const base = Number(target.base_target)
+  if (n > stretch) return '优秀'
+  if (n >= base) return '达标'
+  return '待提升'
+}
+
+function getBarHeight(share) {
+  return Math.round((Number(share) / 1.0) * 160)
+}
+
+function getShortTitle(title) {
+  if (!title) return '?'
+  return title.length > 9 ? title.slice(0, 9) + '...' : title
+}
+
+/**
+ * 计算当年完成率（纯前端，量纲已确认为 0~1 小数直接相除）
+ * - "当年"= 当前自然年（new Date().getFullYear()）
+ * - avg(audience_share) 当年已录期 ÷ base/stretch_target
+ * - 无数据时始终返回 { base: '—', stretch: '—' }
+ * - 完成率可 >100% 正常显示
+ */
+function calcCompletionRate(episodes, targets) {
+  const yearInt = new Date().getFullYear()
+  const yearEpisodes = episodes.filter(e => {
+    const airYear = new Date(e.air_date).getFullYear()
+    return airYear === yearInt && e.audience_share != null
+  })
+  const target = (targets || []).find(t => Number(t.year) === yearInt)
+
+  if (!yearEpisodes.length || !target) return { base: '—', stretch: '—' }
+
+  const avgShare = yearEpisodes.reduce((sum, e) => sum + Number(e.audience_share), 0) / yearEpisodes.length
+  const baseRate = avgShare / Number(target.base_target)
+  const stretchRate = avgShare / Number(target.stretch_target)
+  return {
+    base: (baseRate * 100).toFixed(1) + '%',
+    stretch: (stretchRate * 100).toFixed(1) + '%',
+  }
+}
+
+function Dashboard() {
+  const { user } = useAuthStore()
+  const [episodes, setEpisodes] = useState([])
+  const [targets, setTargets] = useState([])
+  const [cover, setCover] = useState({ cover_path: null, episode_number: null, episode_title: null })
+  const [schedules, setSchedules] = useState([])
+  const [loading, setLoading] = useState(true)
+  const [coverModalOpen, setCoverModalOpen] = useState(false)
+
+  const showChangeCoverBtn = user?.role === 'zhipianren' || user?.role === 'zebian'
+
+  useEffect(() => {
+    Promise.all([
+      listEpisodes(50),
+      listTargets(),
+      getCurrentCover(),
+      getUpcomingSchedules(6),
+    ]).then(([epData, tgtData, coverData, schedData]) => {
+      const sorted = (epData || []).sort((a, b) => new Date(b.air_date) - new Date(a.air_date))
+      setEpisodes(sorted)
+      setTargets(tgtData || [])
+      setCover(coverData || { cover_path: null, episode_number: null, episode_title: null })
+      setSchedules(Array.isArray(schedData) ? schedData : [])
+      setLoading(false)
+    }).catch(() => {
+      setLoading(false)
+    })
+  }, [])
+
+  // 显示最近有份额数据的期次（最多12个）
+  const displayEpisodes = episodes
+    .filter(e => e.audience_share != null)
+    .slice(0, 12)
+
+  const bestEpisode = [...displayEpisodes].sort((a, b) => Number(b.audience_share) - Number(a.audience_share))[0]
+
+  const handleCoverUploaded = (newCover) => {
+    setCover(newCover)
+    setCoverModalOpen(false)
+  }
+
+  // KPI 数据（扩为 5 项：原有 3 项 + 年度完成率 2 项）
+  const completion = calcCompletionRate(episodes, targets)
+  const kpiData = [
+    {
+      icon: <EyeOutlined style={{ color: '#6b8e6b', fontSize: 14 }} />,
+      bg: '#e8f5e9',
+      label: '近12期已录',
+      value: displayEpisodes.length || '--',
+    },
+    {
+      icon: <FireOutlined style={{ color: '#ff9800', fontSize: 14 }} />,
+      bg: '#fff3e0',
+      label: '最佳份额',
+      value: bestEpisode ? (Number(bestEpisode.audience_share) * 100).toFixed(2) + '%' : '--',
+    },
+    {
+      icon: <CalendarOutlined style={{ color: '#2196f3', fontSize: 14 }} />,
+      bg: '#e3f2fd',
+      label: '年度目标',
+      value: targets.length || '--',
+    },
+    {
+      icon: <LineChartOutlined style={{ color: '#7b5e9e', fontSize: 14 }} />,
+      bg: '#f3e5f5',
+      label: '基础目标完成率',
+      value: completion.base,
+    },
+    {
+      icon: <AimOutlined style={{ color: '#c0584f', fontSize: 14 }} />,
+      bg: '#fce4ec',
+      label: '摸高目标完成率',
+      value: completion.stretch,
+    },
+  ]
+
+  return (
+    <div className="dashboard">
+      {/* ===== Banner ===== */}
+      <div className="dashboard-banner">
+        {/* 题图底图 或 默认渐变 */}
+        {cover.cover_path ? (
+          <img
+            src={cover.cover_path}
+            alt="题图"
+            className="banner-bg"
+          />
+        ) : (
+          <div className="banner-default-bg" />
+        )}
+        {/* 暗化渐变层 */}
+        <div className="banner-gradient" />
+        {/* 右上角 Logo */}
+        <img
+          src="/src/assets/junshikeji_logo.png"
+          alt="栏目logo"
+          className="banner-logo"
+          onError={e => { e.target.style.display = 'none' }}
+        />
+        {/* 左侧文字 */}
+        <div className="banner-text">
+          <p className="banner-eyebrow">本月收视最佳·题图</p>
+          <h2 className="banner-title">
+            {cover.episode_number
+              ? `第 ${cover.episode_number} 期`
+              : bestEpisode
+                ? `第 ${bestEpisode.episode_number} 期`
+                : '暂无收视数据'}
+            {cover.episode_title || (bestEpisode ? ` · ${bestEpisode.program_name}` : '')}
+          </h2>
+          <p className="banner-subtitle">
+            {bestEpisode
+              ? `收视份额 ${Number(bestEpisode.audience_share).toFixed(4)} · 当月最高`
+              : '暂无收视数据'}
+          </p>
+          <p className="banner-hint">建议：可选用近期收视表现好的节目海报</p>
+        </div>
+        {/* 右下角换图按钮 */}
+        {showChangeCoverBtn && (
+          <Button
+            className="change-cover-btn"
+            icon={<PictureOutlined />}
+            onClick={() => setCoverModalOpen(true)}
+          >
+            更换题图
+          </Button>
+        )}
+      </div>
+
+      {/* ===== KPI 细条 ===== */}
+      <div className="kpi-strip">
+        {kpiData.map((item, i) => (
+          <div key={i} className="kpi-chip">
+            <div className="kpi-icon" style={{ background: item.bg }}>
+              {item.icon}
+            </div>
+            <div>
+              <p className="kpi-label">{item.label}</p>
+              <p className="kpi-value">{item.value}</p>
+            </div>
+          </div>
+        ))}
+      </div>
+
+      {/* ===== 近9期收视柱图（全宽） ===== */}
+      <div className="bar-chart-card">
+        <div className="bar-chart-header">
+          <h3 className="bar-chart-title">近 12 期收视份额</h3>
+          <span className="bar-chart-hint">悬浮看编导/期次/收视率·颜色只对收视份额</span>
+        </div>
+        {loading ? (
+          <div className="chart-loading"><Spin size="small" /> 加载中...</div>
+        ) : (
+          <>
+            <div className="bars-wrapper">
+              {displayEpisodes.map((ep) => {
+                const airYear = new Date(ep.air_date).getFullYear()
+                const color = ep.audience_share != null
+                  ? getShareColor(ep.audience_share, targets, airYear)
+                  : '#ccc'
+                const colorLabel = ep.audience_share != null
+                  ? getColorLabel(ep.audience_share, targets, airYear)
+                  : '无数据'
+                const tooltipContent = `
+                  <strong>第 ${ep.episode_number} 期 · ${ep.program_name}</strong><br/>
+                  编导：${ep.editor_name_snapshot || '未知'}<br/>
+                  收视份额：${ep.audience_share != null ? Number(ep.audience_share).toFixed(4) : '无数据'}（${colorLabel}）<br/>
+                  收视率：${ep.audience_rating != null ? Number(ep.audience_rating).toFixed(4) : '无数据'}
+                `
+                return (
+                  <div key={ep.id} className="bar-item">
+                    <div className="bar-number" style={{ color }}>
+                      {ep.audience_share != null ? Number(ep.audience_share).toFixed(2) : '--'}
+                    </div>
+                    <Tooltip title={<span dangerouslySetInnerHTML={{ __html: tooltipContent }} />}>
+                      <div
+                        className="bar"
+                        style={{
+                          height: ep.audience_share != null ? getBarHeight(ep.audience_share) : 20,
+                          backgroundColor: color,
+                        }}
+                      />
+                    </Tooltip>
+                    <div className="bar-program" title={ep.program_name}>
+                      {getShortTitle(ep.program_name)}
+                    </div>
+                  </div>
+                )
+              })}
+            </div>
+            <div className="chart-legend">
+              <span className="legend-item">
+                <span className="dot" style={{ background: '#7aa874' }}></span>绿=未达基础目标（待提升）
+              </span>
+              <span className="legend-item">
+                <span className="dot" style={{ background: '#5b8db8' }}></span>蓝=未达摸高目标（达标）
+              </span>
+              <span className="legend-item">
+                <span className="dot" style={{ background: '#c0584f' }}></span>红=超过摸高目标（优秀）
+              </span>
+            </div>
+          </>
+        )}
+      </div>
+
+      {/* ===== 下方两栏 ===== */}
+      <div className="bottom-row">
+        {/* 左：雷达占位 */}
+        <div className="side-card">
+          <h3 className="side-card-title">军事科技热点雷达</h3>
+          <div className="radar-placeholder">
+            <span className="radar-placeholder-label">Phase 4c·占位</span>
+            <div className="radar-sample-items">
+              <div className="radar-sample-item">1. 神舟飞船对接空间站</div>
+              <div className="radar-sample-item">2. 新型无人机集群技术</div>
+            </div>
+            <BarChartOutlined style={{ fontSize: 28, color: '#ccc', marginBottom: 4 }} />
+            <p>真实抓取 Phase 4c 实现</p>
+            <small>编导个人首页核心功能</small>
+          </div>
+        </div>
+
+        {/* 右：排播计划 */}
+        <div className="side-card">
+          <h3 className="side-card-title">未来排播计划</h3>
+          {loading ? (
+            <div style={{ textAlign: 'center', color: '#999', padding: '24px 0' }}><Spin size="small" /></div>
+          ) : schedules.length > 0 ? (
+            <ul className="schedule-list">
+              {schedules.map((item, i) => (
+                <li key={i} className="schedule-item">
+                  <span className="schedule-ep">第{item.episode_number}期</span>
+                  <span className="schedule-name" title={item.program_name}>{item.program_name}</span>
+                  <span className="schedule-date">{item.planned_air_date}</span>
+                  <span className="schedule-editor">{item.editor_name_snapshot || ''}</span>
+                </li>
+              ))}
+            </ul>
+          ) : (
+            <div style={{ textAlign: 'center', color: '#999', padding: '24px 0' }}>
+              <CalendarOutlined style={{ fontSize: 28, color: '#ccc', marginBottom: 8 }} />
+              <p style={{ margin: 0, fontSize: 12 }}>暂无排播数据</p>
+            </div>
+          )}
+        </div>
+      </div>
+
+      {/* 换图 Modal */}
+      <ChangeCoverModal
+        open={coverModalOpen}
+        onClose={() => setCoverModalOpen(false)}
+        onUploaded={handleCoverUploaded}
+        episodes={episodes.slice(0, 20)}
+      />
+    </div>
+  )
+}
+
+export default Dashboard

frontend/src/pages/Doco/Doco.jsx

@@ -0,0 +1,13 @@
+import { Result } from 'antd'
+
+function Doco() {
+  return (
+    <Result
+      status="info"
+      title="文稿对齐模块"
+      subTitle="该模块计划在 1.0 主体上线后作为独立 Phase 开发，详见 doco_project_design.md"
+    />
+  )
+}
+
+export default Doco

frontend/src/pages/EditorDesk/EditorDesk.jsx

@@ -0,0 +1,369 @@
+import { useState, useEffect } from 'react'
+import { Tabs, Table, Button, Modal, Form, Input, InputNumber, DatePicker, message, Space, Upload, Popconfirm } from 'antd'
+import { PlusOutlined, EditOutlined, DeleteOutlined, DownloadOutlined, UploadOutlined } from '@ant-design/icons'
+import { listEpisodes, createEpisode, updateEpisode, deleteEpisode } from '../../services/episodeService'
+import { listTargets, createTarget, updateTarget, deleteTarget } from '../../services/yearlyTargetService'
+import http from '../../services/http'
+
+const { TabPane } = Tabs
+
+// ============================================================
+// Tab1: 节目期次
+// ============================================================
+function EpisodesTab() {
+  const [episodes, setEpisodes] = useState([])
+  const [loading, setLoading] = useState(false)
+  const [modalVisible, setModalVisible] = useState(false)
+  const [editing, setEditing] = useState(null)
+  const [form] = Form.useForm()
+
+  useEffect(() => {
+    loadEpisodes()
+  }, [])
+
+  const loadEpisodes = () => {
+    setLoading(true)
+    listEpisodes().then(data => {
+      setEpisodes((data || []).sort((a, b) => new Date(b.air_date) - new Date(a.air_date)))
+      setLoading(false)
+    }).catch(() => {
+      message.error('加载节目列表失败')
+      setLoading(false)
+    })
+  }
+
+  const handleCreate = () => {
+    setEditing(null)
+    form.resetFields()
+    setModalVisible(true)
+  }
+
+  const handleEdit = (record) => {
+    setEditing(record)
+    form.setFieldsValue({
+      episode_number: record.episode_number,
+      program_name: record.program_name,
+      air_date: record.air_date ? record.air_date : null,
+      editor_name_snapshot: record.editor_name_snapshot || '',
+      audience_share: record.audience_share,
+      audience_rating: record.audience_rating,
+      is_rerun: record.is_rerun,
+      notes: record.notes,
+    })
+    setModalVisible(true)
+  }
+
+  const handleDelete = (id) => {
+    deleteEpisode(id).then(() => {
+      message.success('删除成功')
+      loadEpisodes()
+    }).catch(err => {
+      message.error(err.response?.data?.detail || '删除失败')
+    })
+  }
+
+  const handleSubmit = () => {
+    form.validateFields().then(values => {
+      const payload = {
+        ...values,
+        air_date: values.air_date ? values.air_date.format('YYYY-MM-DD') : null,
+      }
+      const promise = editing
+        ? updateEpisode(editing.id, payload)
+        : createEpisode(payload)
+      promise.then(() => {
+        message.success(editing ? '更新成功' : '创建成功')
+        setModalVisible(false)
+        loadEpisodes()
+      }).catch(err => {
+        message.error(err.response?.data?.detail || '操作失败')
+      })
+    })
+  }
+
+  const columns = [
+    { title: '期次号', dataIndex: 'episode_number', key: 'episode_number', width: 80 },
+    { title: '节目名', dataIndex: 'program_name', key: 'program_name' },
+    { title: '播出日期', dataIndex: 'air_date', key: 'air_date', render: d => d || '-' },
+    { title: '编导', dataIndex: 'editor_name_snapshot', key: 'editor_name_snapshot', render: v => v || '-' },
+    { title: '收视份额', dataIndex: 'audience_share', key: 'audience_share', render: v => v != null ? v.toFixed(4) : '-' },
+    { title: '收视率', dataIndex: 'audience_rating', key: 'audience_rating', render: v => v != null ? v.toFixed(2) : '-' },
+    {
+      title: '操作',
+      key: 'action',
+      render: (_, record) => (
+        <Space>
+          <Button size="small" icon={<EditOutlined />} onClick={() => handleEdit(record)} />
+          <Popconfirm title="确认删除？" onConfirm={() => handleDelete(record.id)}>
+            <Button size="small" danger icon={<DeleteOutlined />} />
+          </Popconfirm>
+        </Space>
+      ),
+    },
+  ]
+
+  return (
+    <div>
+      <div style={{ marginBottom: 16, display: 'flex', justifyContent: 'flex-end' }}>
+        <Button type="primary" icon={<PlusOutlined />} onClick={handleCreate}>新增期次</Button>
+      </div>
+      <Table columns={columns} dataSource={episodes} rowKey="id" loading={loading} pagination={{ pageSize: 20 }} />
+      <Modal title={editing ? '编辑期次' : '新增期次'} open={modalVisible} onOk={handleSubmit} onCancel={() => setModalVisible(false)} okText="保存" width={520}>
+        <Form form={form} layout="vertical" style={{ marginTop: 16 }}>
+          <Form.Item name="episode_number" label="期次号" rules={[{ required: true, message: '请输入期次号' }]}>
+            <InputNumber style={{ width: '100%' }} placeholder="如 1248" disabled={!!editing} />
+          </Form.Item>
+          <Form.Item name="program_name" label="节目名称" rules={[{ required: true, message: '请输入节目名称' }]}>
+            <Input placeholder="如《军事科技》武器解析系列" />
+          </Form.Item>
+          <Form.Item name="air_date" label="播出日期" rules={[{ required: true, message: '请选择播出日期' }]}>
+            <DatePicker style={{ width: '100%' }} format="YYYY-MM-DD" disabled={!!editing} />
+          </Form.Item>
+          <Form.Item name="editor_name_snapshot" label="编导姓名（快照）">
+            <Input placeholder="可留空，离职编导只存姓名" />
+          </Form.Item>
+          <Form.Item name="audience_share" label="收视份额（0-1）">
+            <InputNumber style={{ width: '100%' }} min={0} max={1} precision={4} placeholder="0.72" />
+          </Form.Item>
+          <Form.Item name="audience_rating" label="收视率">
+            <InputNumber style={{ width: '100%' }} min={0} precision={2} placeholder="3.5" />
+          </Form.Item>
+          <Form.Item name="notes" label="备注">
+            <Input.TextArea rows={2} placeholder="可选" />
+          </Form.Item>
+        </Form>
+      </Modal>
+    </div>
+  )
+}
+
+// ============================================================
+// Tab2: 年度目标
+// ============================================================
+function YearlyTargetsTab() {
+  const [targets, setTargets] = useState([])
+  const [loading, setLoading] = useState(false)
+  const [modalVisible, setModalVisible] = useState(false)
+  const [editing, setEditing] = useState(null)
+  const [form] = Form.useForm()
+
+  useEffect(() => {
+    loadTargets()
+  }, [])
+
+  const loadTargets = () => {
+    setLoading(true)
+    listTargets().then(data => {
+      setTargets((data || []).sort((a, b) => b.year - a.year))
+      setLoading(false)
+    }).catch(() => {
+      message.error('加载年度目标失败')
+      setLoading(false)
+    })
+  }
+
+  const handleCreate = () => {
+    setEditing(null)
+    form.resetFields()
+    setModalVisible(true)
+  }
+
+  const handleEdit = (record) => {
+    setEditing(record)
+    form.setFieldsValue({
+      year: record.year,
+      base_target: record.base_target,
+      stretch_target: record.stretch_target,
+      notes: record.notes,
+    })
+    setModalVisible(true)
+  }
+
+  const handleDelete = (id) => {
+    deleteTarget(id).then(() => {
+      message.success('删除成功')
+      loadTargets()
+    }).catch(err => {
+      message.error(err.response?.data?.detail || '删除失败')
+    })
+  }
+
+  const handleSubmit = () => {
+    form.validateFields().then(values => {
+      const promise = editing
+        ? updateTarget(editing.id, values)
+        : createTarget(values)
+      promise.then(() => {
+        message.success(editing ? '更新成功' : '创建成功')
+        setModalVisible(false)
+        loadTargets()
+      }).catch(err => {
+        const detail = err.response?.data?.detail || ''
+        if (detail.includes('已存在') || detail.includes('历史年份')) {
+          message.error('该年目标已存在，历史年份不可覆盖')
+        } else {
+          message.error(detail || '操作失败')
+        }
+      })
+    })
+  }
+
+  const columns = [
+    { title: '年份', dataIndex: 'year', key: 'year', width: 80 },
+    { title: '基础目标', dataIndex: 'base_target', key: 'base_target', render: v => v?.toFixed(4) || '-' },
+    { title: '摸高目标', dataIndex: 'stretch_target', key: 'stretch_target', render: v => v?.toFixed(4) || '-' },
+    { title: '备注', dataIndex: 'notes', key: 'notes', render: v => v || '-' },
+    {
+      title: '操作',
+      key: 'action',
+      render: (_, record) => (
+        <Space>
+          <Button size="small" icon={<EditOutlined />} onClick={() => handleEdit(record)} />
+          <Popconfirm title="确认删除？" onConfirm={() => handleDelete(record.id)}>
+            <Button size="small" danger icon={<DeleteOutlined />} />
+          </Popconfirm>
+        </Space>
+      ),
+    },
+  ]
+
+  return (
+    <div>
+      <div style={{ marginBottom: 16, display: 'flex', justifyContent: 'flex-end' }}>
+        <Button type="primary" icon={<PlusOutlined />} onClick={handleCreate}>新增年度目标</Button>
+      </div>
+      <Table columns={columns} dataSource={targets} rowKey="id" loading={loading} pagination={false} />
+      <Modal title={editing ? '编辑年度目标' : '新增年度目标'} open={modalVisible} onOk={handleSubmit} onCancel={() => setModalVisible(false)} okText="保存">
+        <Form form={form} layout="vertical" style={{ marginTop: 16 }}>
+          <Form.Item
+            name="year"
+            label="年份"
+            rules={[{ required: true, message: '请输入年份' }]}
+          >
+            <InputNumber
+              style={{ width: '100%' }}
+              placeholder="如 2024"
+              min={2000}
+              max={2100}
+              disabled={!!editing}
+            />
+          </Form.Item>
+          <Form.Item name="base_target" label="基础目标" rules={[{ required: true, message: '请输入基础目标' }]}>
+            <InputNumber style={{ width: '100%' }} min={0} max={1} precision={4} placeholder="0.6448" />
+          </Form.Item>
+          <Form.Item name="stretch_target" label="摸高目标" rules={[{ required: true, message: '请输入摸高目标' }]}>
+            <InputNumber style={{ width: '100%' }} min={0} max={1} precision={4} placeholder="0.8989" />
+          </Form.Item>
+          <Form.Item name="notes" label="备注">
+            <Input.TextArea rows={2} placeholder="可选" />
+          </Form.Item>
+        </Form>
+      </Modal>
+    </div>
+  )
+}
+
+// ============================================================
+// Tab3: Excel 批量导入
+// ============================================================
+function ImportTab() {
+  const [uploading, setUploading] = useState(false)
+  const [result, setResult] = useState(null)
+
+  const downloadTemplate = () => {
+    window.open('/api/imports/template', '_blank')
+  }
+
+  const uploadProps = {
+    name: 'file',
+    accept: '.xlsx',
+    showUploadList: false,
+    beforeUpload: (file) => {
+      if (!file.filename.endsWith('.xlsx') && !file.name.endsWith('.xlsx')) {
+        message.error('仅支持 .xlsx 格式')
+        return false
+      }
+      setUploading(true)
+      const formData = new FormData()
+      formData.append('file', file)
+      http.post('/imports/episodes', formData, {
+        headers: { 'Content-Type': 'multipart/form-data' },
+      }).then(res => {
+        setResult(res.data)
+        setUploading(false)
+        if (res.data.errors?.length) {
+          message.warning(`导入完成：成功 ${res.data.imported} 条，失败 ${res.data.errors.length} 条`)
+        } else {
+          message.success(`导入成功：${res.data.imported} 条`)
+        }
+      }).catch(err => {
+        setUploading(false)
+        message.error(err.response?.data?.detail || '上传失败')
+      })
+      return false
+    },
+  }
+
+  const downloadErrors = () => {
+    if (!result?.batch_id) return
+    window.open(`/api/imports/errors/${result.batch_id}`, '_blank')
+  }
+
+  return (
+    <div>
+      <div style={{ display: 'flex', gap: 16, marginBottom: 24 }}>
+        <Button icon={<DownloadOutlined />} onClick={downloadTemplate}>下载导入模板</Button>
+        <Upload {...uploadProps}>
+          <Button type="primary" icon={<UploadOutlined />} loading={uploading}>
+            {uploading ? '上传中...' : '选择 Excel 上传'}
+          </Button>
+        </Upload>
+      </div>
+
+      {result && (
+        <div style={{ background: '#f6f8fa', padding: 16, borderRadius: 8 }}>
+          <p style={{ margin: '0 0 8px', fontWeight: 600 }}>导入结果</p>
+          <p style={{ margin: 0 }}>总计：{result.total} 条</p>
+          <p style={{ margin: '4px 0', color: '#52c41a' }}>成功：{result.imported} 条</p>
+          {result.errors?.length > 0 && (
+            <>
+              <p style={{ margin: '4px 0', color: '#cf1322' }}>失败：{result.errors.length} 条</p>
+              <Button size="small" onClick={downloadErrors} style={{ marginTop: 8 }}>
+                下载失败行
+              </Button>
+            </>
+          )}
+        </div>
+      )}
+
+      <div style={{ marginTop: 24, color: '#999', fontSize: 12 }}>
+        <p>支持批量导入节目期次，导入失败时可根据提示修正后重新上传。</p>
+        <p>注意：Phase 2 不支持重播行导入，有重播行请留空。</p>
+      </div>
+    </div>
+  )
+}
+
+// ============================================================
+// 主组件
+// ============================================================
+function EditorDesk() {
+  return (
+    <div style={{ padding: '0 0 24px' }}>
+      <h2 style={{ margin: '0 0 16px', fontSize: 20, fontWeight: 600 }}>责编录入</h2>
+      <Tabs defaultActiveKey="episodes">
+        <TabPane tab="节目期次" key="episodes">
+          <EpisodesTab />
+        </TabPane>
+        <TabPane tab="年度目标" key="targets">
+          <YearlyTargetsTab />
+        </TabPane>
+        <TabPane tab="Excel 批量导入" key="import">
+          <ImportTab />
+        </TabPane>
+      </Tabs>
+    </div>
+  )
+}
+
+export default EditorDesk

frontend/src/pages/EditorHome/EditorHome.jsx

@@ -0,0 +1,17 @@
+import { Card } from 'antd'
+import { UserOutlined } from '@ant-design/icons'
+
+function EditorHome() {
+  return (
+    <Card style={{ borderRadius: 16 }}>
+      <div style={{ textAlign: 'center', padding: '48px 0', color: '#999' }}>
+        <UserOutlined style={{ fontSize: 48, marginBottom: 16 }} />
+        <h3 style={{ color: '#666' }}>个人首页</h3>
+        <p>模块 H · Phase 4c 实施</p>
+        <small>今日热点雷达 · RSS + AI 摘要</small>
+      </div>
+    </Card>
+  )
+}
+
+export default EditorHome

frontend/src/pages/KnowledgeBase/KnowledgeBase.jsx

@@ -0,0 +1,423 @@
+import { useState, useEffect } from 'react'
+import { Card, Upload, Table, Button, Space, Select, Popconfirm, message, Tag, Input } from 'antd'
+import { DeleteOutlined, ReloadOutlined, UploadOutlined, SearchOutlined, ClearOutlined } from '@ant-design/icons'
+import useAuthStore from '../../stores/authStore'
+import knowledgeService from '../../services/knowledgeService'
+import KnowledgeTree from '../../components/KnowledgeTree/KnowledgeTree'
+
+const { Dragger } = Upload
+
+// source_type 枚举值（固定五类，不写死但供 Select 用）
+const SOURCE_TYPE_OPTIONS = [
+  { label: '全部类型', value: '' },
+  { label: '杂志文章', value: 'military_report' },
+  { label: '节目文稿', value: 'manuscript' },
+  { label: '报题单', value: 'baoti' },
+]
+
+// source_type 中文标签
+const SOURCE_TYPE_LABEL = {
+  military_report: '杂志文章',
+  manuscript: '节目文稿',
+  baoti: '报题单',
+  manual: '其他',
+}
+
+export default function KnowledgeBase() {
+  const { user } = useAuthStore()
+  const [items, setItems] = useState([])
+  const [treeData, setTreeData] = useState([])
+  const [loading, setLoading] = useState(false)
+  const [uploading, setUploading] = useState(false)
+  const [sources, setSources] = useState([]) // 出处下拉选项
+  const [sourceTypeFilter, setSourceTypeFilter] = useState('')
+  const [sourceDetailFilter, setSourceDetailFilter] = useState('')
+  const [selectedTreeNode, setSelectedTreeNode] = useState(null) // { type, detail }
+
+  // 搜索状态
+  const [searchQuery, setSearchQuery] = useState('')
+  const [searchResults, setSearchResults] = useState([])
+  const [searchLoading, setSearchLoading] = useState(false)
+
+  const fetchItems = async () => {
+    setLoading(true)
+    try {
+      const data = await knowledgeService.listItems(sourceTypeFilter || null)
+      setItems(data)
+    } catch {
+      message.error('加载知识库失败')
+    } finally {
+      setLoading(false)
+    }
+  }
+
+  const fetchTree = async () => {
+    try {
+      const data = await knowledgeService.getGroupedItems()
+      setTreeData(data || [])
+    } catch {
+      // ignore
+    }
+  }
+
+  const fetchSources = async () => {
+    try {
+      const data = await knowledgeService.listSources()
+      // 接口返回 [{source: "航空知识 2026年第1期"}, ...]，提取 source 字段
+      setSources(data.map(s => s.source).filter(Boolean))
+    } catch {
+      // ignore
+    }
+  }
+
+  useEffect(() => {
+    fetchItems()
+    fetchTree()
+    fetchSources()
+  }, [sourceTypeFilter])
+
+  // 上传
+  const handleUpload = async (file) => {
+    if (!file.name.endsWith('.md')) {
+      message.warning('仅支持 .md 文件')
+      return false
+    }
+    setUploading(true)
+    try {
+      const result = await knowledgeService.uploadFiles([file])
+      const { uploaded, errors } = result
+      if (uploaded.length > 0) {
+        message.success(`成功入库 ${uploaded.length} 篇`)
+        fetchItems()
+        fetchTree()
+      }
+      if (errors.length > 0) {
+        errors.forEach(e => message.error(`${e.file}: ${e.error}`))
+      }
+    } catch {
+      message.error('上传失败')
+    } finally {
+      setUploading(false)
+    }
+    return false // 阻止默认上传行为
+  }
+
+  // 删除
+  const handleDelete = async (id) => {
+    try {
+      await knowledgeService.deleteItem(id)
+      message.success('删除成功')
+      fetchItems()
+      fetchTree()
+    } catch {
+      message.error('删除失败')
+    }
+  }
+
+  // 搜索处理
+  const handleSearch = async () => {
+    if (!searchQuery.trim()) {
+      setSearchResults([])
+      return
+    }
+    setSearchLoading(true)
+    try {
+      const data = await knowledgeService.searchItems(searchQuery.trim())
+      setSearchResults(data.results || [])
+    } catch {
+      message.error('搜索失败')
+      setSearchResults([])
+    } finally {
+      setSearchLoading(false)
+    }
+  }
+
+  // 清空搜索 → 恢复树形浏览（不丢树状态）
+  const handleClearSearch = () => {
+    setSearchQuery('')
+    setSearchResults([])
+  }
+
+  // 树节点选中 → 联动过滤
+  const handleNodeSelect = (node) => {
+    setSelectedTreeNode(node)
+  }
+
+  // 根据树节点过滤列表
+  const displayedItems = (() => {
+    let result = items
+
+    // 树节点过滤优先（覆盖原有 Select 筛选）
+    if (selectedTreeNode) {
+      const { type, detail } = selectedTreeNode
+      result = result.filter(i => {
+        if (i.source_type !== type) return false
+        if (detail !== null) {
+          // 按大类决定用哪个字段比对（节目文稿=author，杂志文章=source_detail）
+          if (type === 'manuscript') {
+            if (i.author !== detail) return false
+          } else {
+            // 杂志文章等用 source_detail
+            if (i.source_detail !== detail) return false
+          }
+        }
+        return true
+      })
+    } else if (sourceDetailFilter) {
+      // 保留原有的 source_detail 筛选逻辑
+      result = result.filter(i => i.source_detail === sourceDetailFilter)
+    }
+
+    return result
+  })()
+
+  const columns = [
+    {
+      title: '标题',
+      dataIndex: 'title',
+      key: 'title',
+      width: 280,
+      render: (text) => (
+        <span style={{ fontWeight: 500, color: '#3b4a3b', overflow: 'hidden', textOverflow: 'ellipsis', whiteSpace: 'nowrap', display: 'block' }}>{text}</span>
+      ),
+    },
+    {
+      title: '作者',
+      dataIndex: 'author',
+      key: 'author',
+      width: 100,
+      render: (val) => val || '-',
+    },
+    {
+      title: '播出/发表时间',
+      dataIndex: 'publish_date',
+      key: 'publish_date',
+      width: 160,
+      render: (val) => val || '-',
+    },
+    {
+      title: '出处',
+      dataIndex: 'source_detail',
+      key: 'source_detail',
+      width: 200,
+      render: (val) => val || '-',
+    },
+    {
+      title: '类型',
+      dataIndex: 'source_type',
+      key: 'source_type',
+      width: 96,
+      render: (val) => (
+        <Tag color="default" style={{ borderRadius: 8 }}>
+          {SOURCE_TYPE_LABEL[val] || val}
+        </Tag>
+      ),
+    },
+    {
+      title: '操作',
+      key: 'action',
+      width: 90,
+      render: (_, record) => (
+        <Popconfirm
+          title="确认删除"
+          description="删除后不可恢复，该篇及其向量将一并清除"
+          onConfirm={() => handleDelete(record.id)}
+          okText="确认删除"
+          cancelText="取消"
+          placement="left"
+        >
+          <Button size="small" danger icon={<DeleteOutlined />}>
+            删除
+          </Button>
+        </Popconfirm>
+      ),
+    },
+  ]
+
+  return (
+    <div style={{ maxWidth: 1200, padding: '12px 16px' }}>
+      {/* 标题栏 */}
+      <div style={{ marginBottom: 10 }}>
+        <h2 style={{ margin: 0, fontSize: 20, fontWeight: 600, color: '#3b4a3b' }}>知识库</h2>
+        <p style={{ margin: '4px 0 0', fontSize: 13, color: '#888' }}>上传 md 笔记 · 语义检索 · 报题参考</p>
+      </div>
+
+      {/* 上传区（收小为一行按钮） */}
+      <div style={{ display: 'flex', alignItems: 'center', gap: 10, marginBottom: 12 }}>
+        <Dragger
+          accept=".md"
+          multiple
+          showUploadList={false}
+          beforeUpload={handleUpload}
+          disabled={uploading}
+        >
+          <Button icon={<UploadOutlined />} size="middle" style={{ borderRadius: 10 }}>
+            {uploading ? '正在上传并生成向量…' : '上传 md 文件入库'}
+          </Button>
+        </Dragger>
+        <span style={{ fontSize: 12, color: '#aaa' }}>支持批量上传，系统自动解析 yaml frontmatter 并生成语义向量</span>
+      </div>
+
+      {/* 搜索栏（搜索模式/浏览模式互斥，清空恢复树形浏览不丢状态） */}
+      <div style={{ display: 'flex', alignItems: 'center', gap: 8, marginBottom: 12 }}>
+        <Input
+          placeholder="输入一段文字，语义搜索知识库…"
+          value={searchQuery}
+          onChange={e => setSearchQuery(e.target.value)}
+          onPressEnter={handleSearch}
+          style={{ width: 340, borderRadius: 10 }}
+          prefix={<SearchOutlined style={{ color: '#aaa' }} />}
+          suffix={
+            searchQuery ? (
+              <ClearOutlined
+                style={{ color: '#aaa', cursor: 'pointer' }}
+                onClick={handleClearSearch}
+              />
+            ) : null
+          }
+        />
+        <Button
+          icon={<SearchOutlined />}
+          onClick={handleSearch}
+          loading={searchLoading}
+          style={{ borderRadius: 10 }}
+        >
+          搜索
+        </Button>
+        {searchQuery && (
+          <span style={{ fontSize: 12, color: '#888' }}>
+            {searchResults.length > 0
+              ? `找到 ${searchResults.length} 条相关结果`
+              : '无相关结果'}
+          </span>
+        )}
+      </div>
+
+      {/* 搜索结果列表（搜索模式显示，浏览模式隐藏） */}
+      {searchQuery && searchResults.length > 0 && (
+        <div style={{ marginBottom: 12 }}>
+          <div style={{ fontSize: 13, color: '#666', marginBottom: 8, fontWeight: 500 }}>
+            搜索结果
+          </div>
+          {searchResults.map(item => (
+            <Card
+              key={item.id}
+              size="small"
+              style={{ borderRadius: 10, marginBottom: 8 }}
+              bodyStyle={{ padding: '12px 16px' }}
+            >
+              <div style={{ display: 'flex', justifyContent: 'space-between', alignItems: 'flex-start' }}>
+                <div style={{ flex: 1 }}>
+                  <div style={{ fontWeight: 600, fontSize: 14, color: '#3b4a3b', marginBottom: 4 }}>
+                    {item.title}
+                  </div>
+                  <div style={{ display: 'flex', gap: 8, alignItems: 'center', marginBottom: 6 }}>
+                    <Tag color="default" style={{ borderRadius: 6, fontSize: 12 }}>
+                      {SOURCE_TYPE_LABEL[item.source_type] || item.source_type}
+                    </Tag>
+                    {item.author && (
+                      <span style={{ fontSize: 12, color: '#888' }}>作者：{item.author}</span>
+                    )}
+                    {item.source_detail && (
+                      <span style={{ fontSize: 12, color: '#888' }}>出处：{item.source_detail}</span>
+                    )}
+                  </div>
+                  <div
+                    style={{
+                      fontSize: 12,
+                      color: '#555',
+                      background: '#f5f5f5',
+                      borderRadius: 6,
+                      padding: '6px 10px',
+                      lineHeight: 1.6,
+                      overflow: 'hidden',
+                      textOverflow: 'ellipsis',
+                      display: '-webkit-box',
+                      WebkitLineClamp: 3,
+                      WebkitBoxOrient: 'vertical',
+                    }}
+                  >
+                    {item.snippet}
+                  </div>
+                </div>
+                  <div style={{
+                    marginLeft: 16,
+                    minWidth: 56,
+                    textAlign: 'center',
+                    background: item.similarity >= 0.7 ? '#e6f4ff' : '#f5f5f5',
+                    borderRadius: 8,
+                    padding: '4px 8px',
+                  }}>
+                    <div style={{ fontSize: 18, fontWeight: 700, color: item.similarity >= 0.7 ? '#1677ff' : '#666' }}>
+                      {Math.max(0, Math.round(item.similarity * 100))}%
+                    </div>
+                    <div style={{ fontSize: 11, color: '#888' }}>相关度</div>
+                  </div>
+              </div>
+            </Card>
+          ))}
+        </div>
+      )}
+
+      {/* 主体：左侧树 + 右侧列表（flex 左右并列，宽度稳定） */}
+      <div style={{ display: 'flex', gap: 12, alignItems: 'flex-start' }}>
+        {/* 左侧树 */}
+        <Card
+          style={{ borderRadius: 14, width: 280, flexShrink: 0 }}
+          bodyStyle={{ padding: 0, height: '100%' }}
+        >
+          <KnowledgeTree
+            treeData={treeData}
+            onNodeSelect={handleNodeSelect}
+            selectedKey={selectedTreeNode ? `${selectedTreeNode.type}${selectedTreeNode.detail ? '|' + selectedTreeNode.detail : ''}` : 'all'}
+          />
+        </Card>
+
+        {/* 右侧列表 + 筛选栏 */}
+        <div style={{ flex: 1, minWidth: 0 }}>
+          {/* 筛选栏 */}
+          <Card style={{ borderRadius: 14, marginBottom: 12 }}>
+            <Space size="middle" wrap>
+              <span style={{ fontSize: 14, color: '#666' }}>筛选：</span>
+              <Select
+                placeholder="按类型"
+                options={SOURCE_TYPE_OPTIONS}
+                value={sourceTypeFilter}
+                onChange={val => { setSourceTypeFilter(val); setSourceDetailFilter(''); setSelectedTreeNode(null) }}
+                style={{ width: 140 }}
+                allowClear
+              />
+              <Select
+                placeholder="按出处"
+                options={[{ label: '全部出处', value: '' }, ...sources.map(s => ({ label: s, value: s }))]}
+                value={sourceDetailFilter}
+                onChange={val => { setSourceDetailFilter(val); setSelectedTreeNode(null) }}
+                style={{ width: 200 }}
+                allowClear
+              />
+              <Button
+                icon={<ReloadOutlined />}
+                onClick={() => { fetchItems(); fetchTree() }}
+                size="small"
+              >
+                刷新
+              </Button>
+            </Space>
+          </Card>
+
+          {/* 列表 */}
+          <Card style={{ borderRadius: 14 }}>
+            <Table
+              columns={columns}
+              dataSource={displayedItems}
+              rowKey="id"
+              loading={loading}
+              pagination={{ pageSize: 10, size: 'small' }}
+              locale={{ emptyText: '暂无知识库条目，上传 md 文件开始入库' }}
+            />
+          </Card>
+        </div>
+      </div>
+    </div>
+  )
+}

frontend/src/pages/Login/Login.css

@@ -0,0 +1,29 @@
+.login-page {
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  min-height: 100vh;
+  background: var(--color-bg-cream);
+}
+
+.login-card {
+  width: 400px;
+  border-radius: var(--radius-card);
+  box-shadow: 0 4px 24px rgba(0, 0, 0, 0.08);
+}
+
+.login-header {
+  text-align: center;
+  margin-bottom: 32px;
+}
+
+.login-header h1 {
+  font-size: 28px;
+  color: var(--color-primary-green);
+  margin: 0 0 8px;
+}
+
+.login-header p {
+  color: var(--color-text-secondary);
+  margin: 0;
+}

frontend/src/pages/Login/Login.jsx

@@ -0,0 +1,78 @@
+import { useState } from 'react'
+import { Form, Input, Button, Card, message } from 'antd'
+import { UserOutlined, LockOutlined } from '@ant-design/icons'
+import { useNavigate } from 'react-router-dom'
+import useAuthStore from '../../stores/authStore'
+import './Login.css'
+
+function Login() {
+  const [loading, setLoading] = useState(false)
+  const navigate = useNavigate()
+  const { login } = useAuthStore()
+
+  const onFinish = async (values) => {
+    setLoading(true)
+    const result = await login(values.username, values.password)
+    setLoading(false)
+
+    if (result.success) {
+      navigate('/dashboard')
+    } else {
+      message.error(result.error || '登录失败')
+    }
+  }
+
+  return (
+    <div className="login-page">
+      <Card className="login-card" bordered={false}>
+        <div className="login-header">
+          <h1>军事科技</h1>
+          <p>TPS 工作台</p>
+        </div>
+        <Form
+          name="login"
+          onFinish={onFinish}
+          autoComplete="off"
+          layout="vertical"
+          requiredMark={false}
+        >
+          <Form.Item
+            name="username"
+            rules={[{ required: true, message: '请输入用户名' }]}
+          >
+            <Input
+              prefix={<UserOutlined />}
+              placeholder="用户名"
+              size="large"
+            />
+          </Form.Item>
+
+          <Form.Item
+            name="password"
+            rules={[{ required: true, message: '请输入密码' }]}
+          >
+            <Input.Password
+              prefix={<LockOutlined />}
+              placeholder="密码"
+              size="large"
+            />
+          </Form.Item>
+
+          <Form.Item>
+            <Button
+              type="primary"
+              htmlType="submit"
+              size="large"
+              block
+              loading={loading}
+            >
+              登录
+            </Button>
+          </Form.Item>
+        </Form>
+      </Card>
+    </div>
+  )
+}
+
+export default Login

frontend/src/pages/TPS/TPS.jsx

@@ -0,0 +1,17 @@
+import { Card } from 'antd'
+import { FileTextOutlined } from '@ant-design/icons'
+
+function TPS() {
+  return (
+    <Card style={{ borderRadius: 16 }}>
+      <div style={{ textAlign: 'center', padding: '48px 0', color: '#999' }}>
+        <FileTextOutlined style={{ fontSize: 48, marginBottom: 16 }} />
+        <h3 style={{ color: '#666' }}>TPS 选题策划</h3>
+        <p>模块 A / B / C / D / G · Phase 4a 实施</p>
+        <small>查重 + 知识库参考 + 报题单生成</small>
+      </div>
+    </Card>
+  )
+}
+
+export default TPS

frontend/src/pages/UserManage/UserManage.jsx

@@ -0,0 +1,187 @@
+import { useState, useEffect } from 'react'
+import { Table, Button, Modal, Form, Input, Select, Switch, message, Space } from 'antd'
+import { UserAddOutlined, EditOutlined, StopOutlined } from '@ant-design/icons'
+import { listUsers, createUser, updateUser, deactivateUser } from '../../services/userService'
+
+const roleOptions = [
+  { value: 'zhipianren', label: '制片人' },
+  { value: 'zebian', label: '责编' },
+  { value: 'biandao', label: '编导' },
+]
+
+function UserManage() {
+  const [users, setUsers] = useState([])
+  const [loading, setLoading] = useState(false)
+  const [modalVisible, setModalVisible] = useState(false)
+  const [editingUser, setEditingUser] = useState(null)
+  const [form] = Form.useForm()
+
+  useEffect(() => {
+    loadUsers()
+  }, [])
+
+  const loadUsers = () => {
+    setLoading(true)
+    listUsers().then(data => {
+      setUsers(data || [])
+      setLoading(false)
+    }).catch(() => {
+      message.error('加载用户列表失败')
+      setLoading(false)
+    })
+  }
+
+  const handleCreate = () => {
+    setEditingUser(null)
+    form.resetFields()
+    setModalVisible(true)
+  }
+
+  const handleEdit = (record) => {
+    setEditingUser(record)
+    form.setFieldsValue({
+      display_name: record.display_name,
+      role: record.role,
+      profile_text: record.profile_text,
+      is_active: record.is_active,
+    })
+    setModalVisible(true)
+  }
+
+  const handleDeactivate = (record) => {
+    Modal.confirm({
+      title: '停用用户',
+      content: `确认停用用户「${record.display_name}」（${record.username}）？停用后可重新启用。`,
+      okText: '确认停用',
+      cancelText: '取消',
+      onOk: () => deactivateUser(record.id).then(() => {
+        message.success('已停用')
+        loadUsers()
+      }).catch(err => {
+        message.error(err.response?.data?.detail || '停用失败')
+      }),
+    })
+  }
+
+  const handleSubmit = () => {
+    form.validateFields().then(values => {
+      if (editingUser) {
+        updateUser(editingUser.id, values).then(() => {
+          message.success('更新成功')
+          setModalVisible(false)
+          loadUsers()
+        }).catch(err => {
+          message.error(err.response?.data?.detail || '更新失败')
+        })
+      } else {
+        createUser(values).then(() => {
+          message.success('创建成功')
+          setModalVisible(false)
+          loadUsers()
+        }).catch(err => {
+          message.error(err.response?.data?.detail || '创建失败')
+        })
+      }
+    })
+  }
+
+  const columns = [
+    {
+      title: '用户名',
+      dataIndex: 'username',
+      key: 'username',
+    },
+    {
+      title: '显示名',
+      dataIndex: 'display_name',
+      key: 'display_name',
+    },
+    {
+      title: '角色',
+      dataIndex: 'role',
+      key: 'role',
+      render: role => roleOptions.find(o => o.value === role)?.label || role,
+    },
+    {
+      title: '状态',
+      dataIndex: 'is_active',
+      key: 'is_active',
+      render: (active, record) => (
+        <Switch
+          checked={active}
+          onChange={() => handleDeactivate(record)}
+          checkedChildren="在职"
+          unCheckedChildren="已停"
+        />
+      ),
+    },
+    {
+      title: '操作',
+      key: 'action',
+      render: (_, record) => (
+        <Space>
+          <Button size="small" icon={<EditOutlined />} onClick={() => handleEdit(record)}>
+            编辑
+          </Button>
+          {record.is_active && (
+            <Button size="small" danger icon={<StopOutlined />} onClick={() => handleDeactivate(record)}>
+              停用
+            </Button>
+          )}
+        </Space>
+      ),
+    },
+  ]
+
+  return (
+    <div style={{ padding: '0 0 24px' }}>
+      <div style={{ marginBottom: 16, display: 'flex', justifyContent: 'space-between', alignItems: 'center' }}>
+        <h2 style={{ margin: 0, fontSize: 20, fontWeight: 600 }}>用户管理</h2>
+        <Button type="primary" icon={<UserAddOutlined />} onClick={handleCreate}>
+          新建用户
+        </Button>
+      </div>
+
+      <Table
+        columns={columns}
+        dataSource={users}
+        rowKey="id"
+        loading={loading}
+        pagination={{ pageSize: 20 }}
+      />
+
+      <Modal
+        title={editingUser ? '编辑用户' : '新建用户'}
+        open={modalVisible}
+        onOk={handleSubmit}
+        onCancel={() => setModalVisible(false)}
+        okText="保存"
+        cancelText="取消"
+      >
+        <Form form={form} layout="vertical" style={{ marginTop: 16 }}>
+          {!editingUser && (
+            <Form.Item name="username" label="用户名" rules={[{ required: true, message: '请输入用户名' }]}>
+              <Input placeholder="登录用户名" />
+            </Form.Item>
+          )}
+          <Form.Item name="display_name" label="显示名" rules={[{ required: true, message: '请输入显示名' }]}>
+            <Input placeholder="界面显示名称" />
+          </Form.Item>
+          {!editingUser && (
+            <Form.Item name="password" label="密码" rules={[{ required: true, message: '请输入密码' }]}>
+              <Input.Password placeholder="初始密码" />
+            </Form.Item>
+          )}
+          <Form.Item name="role" label="角色" rules={[{ required: true, message: '请选择角色' }]}>
+            <Select placeholder="选择角色" options={roleOptions} />
+          </Form.Item>
+          <Form.Item name="profile_text" label="个人简介">
+            <Input.TextArea rows={3} placeholder="可选" />
+          </Form.Item>
+        </Form>
+      </Modal>
+    </div>
+  )
+}
+
+export default UserManage

frontend/src/services/authService.js

@@ -0,0 +1,18 @@
+import http from './http'
+
+export async function login(username, password) {
+  const response = await http.post('/auth/login', { username, password })
+  return response.data
+}
+
+export async function logout() {
+  const response = await http.post('/auth/logout')
+  return response.data
+}
+
+export async function me() {
+  const response = await http.get('/auth/me')
+  return response.data
+}
+
+export default { login, logout, me }

frontend/src/services/dashboardService.js

@@ -0,0 +1,38 @@
+import http from './http'
+
+/**
+ * 获取当前题图设置
+ * @returns {Promise<{cover_path: string|null, episode_number: number|null, episode_title: string|null}>}
+ */
+export async function getCurrentCover() {
+  const response = await http.get('/dashboard/cover')
+  return response.data
+}
+
+/**
+ * 上传题图（仅制片人/责编可用）
+ * @param {File} file - 海报图片文件
+ * @param {number} episodeNumber - 关联期次号
+ * @param {string} episodeTitle - 关联期次节目名
+ * @returns {Promise<{success: boolean, cover_path: string}>}
+ */
+export async function uploadCover(file, episodeNumber, episodeTitle) {
+  const formData = new FormData()
+  formData.append('file', file)
+  formData.append('episode_number', String(episodeNumber))
+  formData.append('episode_title', episodeTitle)
+  const response = await http.post('/dashboard/cover', formData, {
+    headers: { 'Content-Type': 'multipart/form-data' },
+  })
+  return response.data
+}
+
+/**
+ * 获取未来排播计划
+ * @param {number} limit - 返回条数，默认6
+ * @returns {Promise<Array<{episode_number, program_name, planned_air_date, editor_name_snapshot}>>}
+ */
+export async function getUpcomingSchedules(limit = 6) {
+  const response = await http.get('/schedules/upcoming', { params: { limit } })
+  return response.data
+}

frontend/src/services/episodeService.js

@@ -0,0 +1,21 @@
+import http from './http'
+
+export async function listEpisodes(limit = 50, offset = 0) {
+  const response = await http.get('/episodes', { params: { limit, offset } })
+  return response.data
+}
+
+export async function createEpisode(data) {
+  const response = await http.post('/episodes', data)
+  return response.data
+}
+
+export async function updateEpisode(id, data) {
+  const response = await http.patch(`/episodes/${id}`, data)
+  return response.data
+}
+
+export async function deleteEpisode(id) {
+  const response = await http.delete(`/episodes/${id}`)
+  return response.data
+}

frontend/src/services/http.js

@@ -0,0 +1,8 @@
+import axios from 'axios'
+
+const http = axios.create({
+  baseURL: '/api',
+  withCredentials: true,
+})
+
+export default http

frontend/src/services/knowledgeService.js

@@ -0,0 +1,70 @@
+/**
+ * 知识库 API 服务
+ */
+
+import http from './http'
+
+const knowledgeService = {
+  /**
+   * 上传 md 文件（单个或多个）
+   * @param {File[]} files - File 对象数组
+   */
+  async uploadFiles(files) {
+    const formData = new FormData()
+    files.forEach(f => formData.append('files', f))
+    const resp = await http.post('/knowledge/upload', formData, {
+      headers: { 'Content-Type': 'multipart/form-data' },
+    })
+    return resp.data
+  },
+
+  /**
+   * 获取知识库条目列表
+   * @param {string|null} sourceType - 可选，按 source_type 筛选
+   */
+  async listItems(sourceType = null) {
+    const params = sourceType ? { source_type: sourceType } : {}
+    const resp = await http.get('/knowledge/items', { params })
+    return resp.data
+  },
+
+  /**
+   * 删除知识库条目
+   * @param {number} id
+   */
+  async deleteItem(id) {
+    const resp = await http.delete(`/knowledge/items/${id}`)
+    return resp.data
+  },
+
+  /**
+   * 获取所有不重复的具体出处（供筛选下拉）
+   */
+  async listSources() {
+    const resp = await http.get('/knowledge/sources')
+    return resp.data
+  },
+
+  /**
+   * 获取按来源分组的树形结构（含「全部」根节点）
+   */
+  async getGroupedItems() {
+    const resp = await http.get('/knowledge/grouped')
+    return resp.data
+  },
+
+  /**
+   * 语义检索：输入一段文字，返回最相关的知识库条目
+   * @param {string} queryText - 查询文字
+   * @param {number} topK - 返回条数，默认 5
+   */
+  async searchItems(queryText, topK = 5) {
+    const resp = await http.post('/knowledge/search', {
+      query: queryText,
+      top_k: topK,
+    })
+    return resp.data
+  },
+}
+
+export default knowledgeService

frontend/src/services/userService.js

@@ -0,0 +1,21 @@
+import http from './http'
+
+export async function listUsers() {
+  const response = await http.get('/users')
+  return response.data
+}
+
+export async function createUser(data) {
+  const response = await http.post('/users', data)
+  return response.data
+}
+
+export async function updateUser(id, data) {
+  const response = await http.patch(`/users/${id}`, data)
+  return response.data
+}
+
+export async function deactivateUser(id) {
+  const response = await http.delete(`/users/${id}`)
+  return response.data
+}

frontend/src/services/yearlyTargetService.js

@@ -0,0 +1,21 @@
+import http from './http'
+
+export async function listTargets() {
+  const response = await http.get('/yearly_targets')
+  return response.data
+}
+
+export async function createTarget(data) {
+  const response = await http.post('/yearly_targets', data)
+  return response.data
+}
+
+export async function updateTarget(id, data) {
+  const response = await http.patch(`/yearly_targets/${id}`, data)
+  return response.data
+}
+
+export async function deleteTarget(id) {
+  const response = await http.delete(`/yearly_targets/${id}`)
+  return response.data
+}