From 743d69efd0d0da8a1bd02ae8b3f9dca393f43bdf Mon Sep 17 00:00:00 2001
From: rain-bus <rainandbus@gmail.com>
Date: Sat, 13 Jun 2026 18:31:43 +0800
Subject: [PATCH] refactor: extract admin business logic to services, introduce
 job queue, add derived index helpers

- Move DB operations from routes/admin.py to services/admin.py (get_logs_context, query_summary_statuses, retry_failed, delete/reset operations)
- Add services/jobs.py with Job/JobEvent-based async job queue (create_job, run_job, enqueue_job)
- Add services/derived.py with FTS5 reindex and paper index deletion helpers
- Refactor scheduler to use job queue instead of direct pipeline calls
- Add heartbeat_at/expires_at to TaskLock for lock health tracking
- Remove DESIGN_REVIEW.md
- Update tests: remove redundant integration tests, add unit tests for new services
---
 .env.example                      |   2 +-
 DESIGN_REVIEW.md                  | 468 ------------------------------
 app/cli.py                        |  72 ++++-
 app/database.py                   |  16 +
 app/main.py                       |   7 +
 app/models.py                     |  59 ++++
 app/routes/admin.py               | 383 +++++++-----------------
 app/services/admin.py             | 327 ++++++++++++++++++++-
 app/services/crawler.py           |  19 +-
 app/services/derived.py           | 140 +++++++++
 app/services/jobs.py              | 244 ++++++++++++++++
 app/services/pipeline.py          |  14 +-
 app/services/scheduler.py         |  14 +-
 app/services/summary_persister.py |  19 +-
 tests/test_admin.py               | 239 +++++++++------
 tests/test_crawler.py             |  60 ++++
 tests/test_derived.py             |  80 +++++
 tests/test_jobs.py                | 111 +++++++
 tests/test_pages.py               | 111 -------
 tests/test_searcher.py            |  69 -----
 20 files changed, 1391 insertions(+), 1063 deletions(-)
 delete mode 100644 DESIGN_REVIEW.md
 create mode 100644 app/services/derived.py
 create mode 100644 app/services/jobs.py
 create mode 100644 tests/test_derived.py
 create mode 100644 tests/test_jobs.py

diff --git a/.env.example b/.env.example
index 8ba3f16..d7a619c 100644
--- a/.env.example
+++ b/.env.example
@@ -42,7 +42,7 @@ DATABASE_URL=sqlite:///data/db/papers.db
 # ─── 语义搜索 ─────────────────────────────
 CHROMA_ENABLED=false
 CHROMA_DIR=data/chroma
-EMBED_API_BASE=https://api.siliconflow.cn/v1/embeddings
+EMBED_API_BASE=https://api.siliconflow.cn
 EMBED_API_KEY=your_api_key_here
 EMBED_MODEL=Qwen/Qwen3-Embedding-4B
 EMBED_DIMENSIONS=2560
diff --git a/DESIGN_REVIEW.md b/DESIGN_REVIEW.md
deleted file mode 100644
index 18bb9b6..0000000
--- a/DESIGN_REVIEW.md
+++ /dev/null
@@ -1,468 +0,0 @@
-# 项目设计审查与优化建议
-
-本文档汇总对当前项目的系统设计、流程设计和代码结构的审查结论。重点不在局部代码风格，而在后续稳定运行、失败恢复、数据一致性、可维护性和扩展性。
-
-## 总体评价
-
-项目当前结构整体清晰：FastAPI 路由层较薄，主要业务放在 `app/services/`；抓取、总结、搜索、个人数据、管理后台等能力已有模块化拆分；SQLite + FTS5 对本地个人论文导览站是合理选择。
-
-主要风险集中在以下几类：
-
-- 长任务生命周期没有统一抽象。
-- Pipeline 不是显式状态机，阶段结果难以恢复和追踪。
-- 数据库、文件系统、FTS5、ChromaDB 之间存在多处手工同步。
-- Web、CLI、Scheduler 三个入口的业务策略存在分叉风险。
-- 失败恢复和可观测性不足。
-- 部分同步阻塞任务混入 async Web 流程。
-
-## 高优先级问题
-
-### 1. 缺少统一的任务系统
-
-当前有三套入口都能触发重任务：
-
-- Web 管理后台：`/admin/trigger-pipeline`、`/admin/summarize`
-- APScheduler：每日自动 pipeline
-- CLI：`app.cli crawl/summarize`
-
-它们共享部分 service，但没有统一的“任务创建、排队、运行、重试、取消、状态查询、失败恢复”模型。
-
-影响：
-
-- Web 请求会长时间挂住。
-- CLI、Web、Scheduler 的行为可能逐渐分叉。
-- 任务卡死后只能看到粗粒度的 running lock。
-- 前端难以展示阶段进度。
-- 后续增加重跑、取消、恢复、并发限制会越来越复杂。
-
-建议：
-
-引入统一 Job 模型：
-
-```text
-入口层 -> 创建 Job -> Worker 执行 Job -> JobEvent 记录进度 -> 页面/API 查询状态
-```
-
-建议任务类型：
-
-- `crawl_daily`
-- `summarize_one`
-- `summarize_batch`
-- `pipeline_daily`
-- `refresh_upvotes`
-- `delete_range`
-- `reindex_fts`
-- `reindex_chroma`
-- `extract_images`
-
-建议表结构方向：
-
-- `jobs`: 任务主记录，包含 `id/type/status/owner/created_at/started_at/completed_at/error`
-- `job_events`: 阶段事件，包含 `job_id/stage/status/message/payload_json/created_at`
-- `paper_processing_runs`: 单篇论文处理记录，适合跟踪总结、图片提取、索引状态
-
-### 2. Pipeline 不是显式状态机
-
-当前 pipeline 基本是线性函数：
-
-```text
-crawl -> summarize -> cleanup
-```
-
-阶段输出没有被建模成可查询、可重放、可补偿的状态。
-
-影响：
-
-- crawl 成功但 summarize 部分失败时，整体状态表达不够精确。
-- cleanup 失败是否影响 pipeline 成败语义不清。
-- “今天无数据回退昨天”是隐式逻辑，不是可配置策略。
-- 图片提取、ChromaDB 索引失败被吞掉，用户无法知道哪些增强能力缺失。
-
-建议：
-
-将 pipeline 改为显式阶段状态：
-
-```text
-created
--> crawling
--> crawled
--> summarizing
--> summarized | summarized_partial
--> postprocessing
--> completed | failed | cancelled
-```
-
-每个阶段记录：
-
-- 输入参数
-- 输出统计
-- 错误类型
-- 错误摘要
-- 开始/结束时间
-- 是否可重试
-
-这样可以支持只重跑失败阶段，例如只重跑 ChromaDB 索引、只重跑图片提取、只重跑失败论文总结。
-
-### 3. 数据生命周期没有清晰分层
-
-项目中至少有四类数据：
-
-- 主数据：`papers`、`authors`、`tags`、`summaries`、用户笔记/收藏/阅读状态
-- 派生索引：FTS5、ChromaDB
-- 长期资产：`data/papers/{arxiv_id}` 下的 summary、raw output、images
-- 临时资产：`data/tmp/{arxiv_id}` 下的 PDF、源码、中间文件
-
-现在这些数据由不同流程手动维护。删除、总结、重建索引、清理临时目录都分散在不同模块中。
-
-影响：
-
-- DB 与文件系统可能不一致。
-- DB 与 FTS5/ChromaDB 可能不一致。
-- 删除流程中一部分资源删除成功、一部分失败时难以恢复。
-- 很难判断某篇论文是否“完整可用”。
-
-建议：
-
-明确数据权威源：
-
-- DB 是权威源。
-- FTS5、ChromaDB、summary.json、raw_output.txt、images 都应视为可重建派生物。
-- 删除主数据优先保证 DB 状态正确，派生物清理可以异步补偿。
-- 提供派生数据重建任务。
-
-建议增加命令或任务：
-
-```text
-rebuild-derived --fts
-rebuild-derived --chroma
-rebuild-derived --files
-rebuild-derived --images
-```
-
-### 4. 数据库与文件系统存在双写一致性风险
-
-总结持久化流程会写文件、写 DB、更新状态、提取图片、写 ChromaDB。任一步失败都可能留下不一致状态。
-
-典型风险：
-
-- 文件已写入，但 DB 未提交。
-- DB 标记 done，但 summary.json 缺失。
-- DB done，但图片未提取或 ChromaDB 未索引。
-- 重新总结时旧图片或旧 figures 残留。
-
-建议：
-
-- 将结构化总结以 DB 为准，JSON 文件作为导出/缓存。
-- 文件写入使用临时文件加原子 rename。
-- DB 中记录派生物状态，例如：
-  - `summary_persisted_at`
-  - `fts_indexed_at`
-  - `chroma_indexed_at`
-  - `images_extracted_at`
-  - `derived_error`
-- 提供一致性检查任务，扫描 DB 与文件/索引差异并修复。
-
-### 5. 失败恢复设计偏弱
-
-当前失败主要依赖：
-
-- `summary_status.status`
-- `retry_count`
-- `task_locks`
-- `crawl_logs`
-
-但对以下情况支持不足：
-
-- 进程崩溃后 `processing` 永久残留。
-- `task_locks` 中 running lock 永久残留。
-- 任务执行到一半，部分文件或索引已经写入。
-- 某些后处理失败但主任务显示成功。
-
-建议：
-
-- `task_locks` 增加 `heartbeat_at` 或 `expires_at`。
-- 应用启动时扫描 stale running task，标记为 `failed/stale`。
-- `summary_status` 增加更细阶段字段，例如：
-  - `downloading_pdf`
-  - `generating`
-  - `validating`
-  - `persisting`
-  - `extracting_images`
-  - `indexing`
-- 区分主流程失败和派生增强失败。
-
-### 6. 长任务直接跑在 Web 请求里
-
-管理端触发 pipeline、批量总结、刷新 upvotes、删除数据时，会在请求生命周期内直接执行重任务。
-
-影响：
-
-- HTTP 请求容易超时。
-- 用户关闭页面后任务状态不清晰。
-- Web 进程被重任务占用。
-- 后续很难做取消、进度展示和失败恢复。
-
-建议：
-
-- Web 只创建 job 并返回 `task_id`。
-- 后台 worker 执行任务。
-- 管理页面轮询 `/admin/jobs/{id}` 或使用 SSE 展示进度。
-
-## 中优先级问题
-
-### 7. FTS5 索引手工维护，容易漂移
-
-`papers_fts` 是独立虚拟表。插入论文、更新总结、删除论文时手动同步。
-
-影响：
-
-- 已有论文的 title、abstract、authors、tags 更新时可能不同步。
-- AI 标签新增后，如果没有统一重建，搜索结果可能缺字段。
-- 删除或回滚失败后索引可能残留。
-
-建议：
-
-- 封装统一的 `reindex_paper(db, paper_id)`。
-- 所有修改论文、标签、总结的路径最后调用它。
-- 提供全量 `reindex_fts` 任务。
-- 或用 SQLite trigger 维护基础字段，但总结字段仍建议通过统一函数更新。
-
-### 8. ChromaDB 索引缺少系统级一致性策略
-
-ChromaDB 当前是可选增强能力，索引失败会被日志吞掉，不影响总结主流程。
-
-影响：
-
-- 语义搜索可用性不透明。
-- 某些论文 DB done，但未进入 ChromaDB。
-- 删除或重建时可能出现残留索引。
-
-建议：
-
-- 将 ChromaDB 明确视为派生索引。
-- DB 中记录 `chroma_indexed_at` 和 `chroma_error`。
-- 提供 `reindex_chroma` 任务。
-- 搜索页或管理后台展示语义索引健康状态。
-
-### 9. 同步阻塞操作混入 async 主流程
-
-部分 async 函数内部调用同步阻塞操作，例如 PDF 下载使用 `requests`，图片提取和 PDF 解析也都是同步重任务。
-
-影响：
-
-- FastAPI event loop 可能被阻塞。
-- 多篇并发总结时吞吐不可控。
-- Web 服务响应受后台任务影响。
-
-建议：
-
-- 最佳方案：重任务移到 worker 进程。
-- 过渡方案：对同步阻塞段使用 `asyncio.to_thread()`。
-- PDF 下载统一使用 `httpx.AsyncClient` 或明确放入同步 worker。
-
-### 10. 调度策略和业务流程耦合
-
-Scheduler 固定每日 pipeline 加 30 分钟 upvote refresh；pipeline 内部固定“今天无数据回退昨天”。
-
-影响：
-
-- 策略不易测试和调整。
-- CLI、Web、Scheduler 可能各自实现不同 fallback。
-- 后续支持日期范围、补抓、只总结新增论文会变复杂。
-
-建议将策略配置化：
-
-- `crawl_target_policy`: `today` / `today_then_yesterday` / `date_range`
-- `summarize_scope`: `new_only` / `pending_and_failed` / `date_range`
-- `cleanup_policy`: `after_success` / `always` / `manual`
-- `upvote_refresh_window_days`
-- `pipeline_on_partial_failure`: `continue` / `stop`
-
-### 11. 内嵌 APScheduler 对部署形态敏感
-
-当前调度器嵌入 Web 进程，且多 worker 时只是打印警告。
-
-影响：
-
-- 多 worker、多进程、reload 模式下可能重复或漏跑任务。
-- Web 进程重启会影响调度可靠性。
-- 调度状态和任务执行状态耦合。
-
-建议：
-
-- 本地单机可以保留内嵌调度器。
-- 长期运行建议拆为独立 scheduler 进程。
-- scheduler 只负责创建 job，不直接执行重任务。
-- Web 管理后台只展示 scheduler/job 状态。
-
-### 12. 运行时迁移能力不足
-
-当前 `_migrate()` 只支持补列。
-
-影响：
-
-- 无法可靠处理索引、约束、字段改名、数据回填。
-- 数据库结构演进不可追踪。
-- 生产或长期本地数据升级风险高。
-
-建议：
-
-- 引入 Alembic。
-- 将 FTS5、索引、部分约束和数据回填纳入迁移脚本。
-- 启动时不要静默做复杂迁移，改为显式执行迁移命令。
-
-## 低到中优先级问题
-
-### 13. 配置校验不足
-
-当前配置字段主要是裸类型，缺少合法值和组合校验。
-
-风险示例：
-
-- `SUMMARY_BACKEND` 填错。
-- `SUMMARY_PDF_MODE` 填错。
-- `SCHEDULE_MINUTE + 30` 超过 59。
-- `APP_WORKERS > 1` 且 `SCHEDULER_ENABLED=true`。
-- `CHROMA_ENABLED=true` 但 embedding 配置缺失。
-
-建议：
-
-- 使用 Pydantic validator 校验枚举值和组合条件。
-- 对危险组合启动时报错，而不是只 warning。
-
-### 14. 私有函数跨模块调用说明模块边界不稳定
-
-例如 `summarizer.py` 调用 `_generate_with_retry`、`_persist_summary` 等下划线函数。
-
-影响：
-
-- 模块公开边界不清晰。
-- 后续重构容易误伤。
-- 测试和替换后端不方便。
-
-建议：
-
-- 为 summary 生成、持久化、后处理定义明确公开 API。
-- 下划线函数保留为模块内部实现细节。
-- 引入更明确的 service 类或函数边界，例如：
-  - `SummaryGenerator.generate()`
-  - `SummaryRepository.save()`
-  - `DerivedAssetBuilder.extract_images()`
-
-### 15. 外部依赖缺少统一 adapter
-
-项目依赖多个外部系统：
-
-- HuggingFace API
-- arXiv PDF/source 下载
-- `pi` CLI
-- `claude` CLI
-- Embedding API
-- ChromaDB
-- ONNX layout detector
-
-建议：
-
-- 为每类外部依赖定义 adapter。
-- 统一 timeout、retry、error classification、metrics。
-- 测试中替换 adapter，而不是 patch 深层函数。
-
-### 16. 删除流程事务边界不理想
-
-删除流程中同时删除 FTS5、ChromaDB、本地文件、临时文件、ORM 数据。
-
-影响：
-
-- 文件删除成功后 DB rollback，会造成数据仍在但文件丢失。
-- DB 删除成功后 ChromaDB 删除失败，会造成残留索引。
-- 单篇失败时 rollback 可能影响之前 flush 的状态，逻辑不直观。
-
-建议：
-
-- 删除主数据与删除派生物分两阶段。
-- DB 删除成功后创建派生清理 job。
-- 派生物清理失败可重试，不影响主数据一致性。
-
-## 建议目标架构
-
-如果项目定位是个人本地工具，可以采用轻量目标架构：
-
-```text
-FastAPI Web
-  - 页面和 API
-  - 管理后台
-  - 创建 job
-  - 查询 job 状态
-
-Scheduler
-  - 按策略创建 job
-  - 不直接跑重任务
-
-Worker
-  - crawl
-  - summarize
-  - PDF download
-  - image extraction
-  - FTS/Chroma reindex
-  - cleanup/delete
-
-Storage/Repository
-  - DB 权威数据
-  - 文件资产管理
-  - 派生索引重建
-  - 一致性检查
-```
-
-如果暂时不想引入独立 worker，也可以先在单进程内实现 Job 表和后台任务执行器。这样至少能统一任务状态和恢复机制。
-
-## 推荐实施顺序
-
-### 第一阶段：先解决任务与状态
-
-目标：降低长任务不可控风险。
-
-- 新增 `jobs` 和 `job_events` 表。
-- Web 管理动作改为创建 job 并返回 task_id。
-- Scheduler 改为创建 job。
-- CLI 复用同一套 job runner。
-- 加 stale running job 恢复逻辑。
-
-### 第二阶段：统一派生数据管理
-
-目标：降低 DB、文件、FTS、Chroma 不一致风险。
-
-- 明确 DB 为权威源。
-- 封装 `reindex_paper()`。
-- 增加 `reindex_fts`、`reindex_chroma` 任务。
-- 增加派生数据状态字段或健康检查。
-- 删除流程改为 DB 主删除 + 派生清理 job。
-
-### 第三阶段：改造 pipeline 为状态机
-
-目标：让任务可恢复、可补偿、可观测。
-
-- 拆分 pipeline stages。
-- 每个 stage 记录输入/输出/错误/耗时。
-- 支持从失败阶段重跑。
-- 将 fallback 策略配置化。
-
-### 第四阶段：提升运行可靠性
-
-目标：长期运行更稳。
-
-- 引入 Alembic。
-- 配置加 validator。
-- 同步阻塞操作移出 Web event loop。
-- 外部依赖 adapter 化。
-- 管理后台展示任务、派生索引、失败类型、耗时统计。
-
-## 最小可行改造方案
-
-如果只做最小但收益最大的改动，建议优先做这三项：
-
-1. 增加统一 Job 表和 job runner。
-2. DB 作为权威源，FTS/Chroma/文件全部按派生物处理。
-3. 增加 stale task 恢复和派生数据重建命令。
-
-这三项能显著降低后续复杂度，也不会强迫项目马上拆成多个服务。
-
diff --git a/app/cli.py b/app/cli.py
index 28f8e06..8049d59 100644
--- a/app/cli.py
+++ b/app/cli.py
@@ -26,7 +26,7 @@ def crawl(
     from app.database import SessionLocal, engine
     from app.database import init_db as _init
     from app.models import Paper
-    from app.services.crawler import crawl_daily
+    from app.services.jobs import create_job, run_job
     from app.utils import today_str, yesterday_str
     from sqlalchemy import func, select
 
@@ -55,7 +55,13 @@ def crawl(
                 return
 
         typer.echo(f"📡 开始抓取 {target} ...")
-        result = asyncio.run(crawl_daily(db, target, top_n))
+        job = create_job(
+            db,
+            "crawl_daily",
+            owner="cli_crawl",
+            payload={"target_date": target, "top_n": top_n},
+        )
+        result = asyncio.run(run_job(db, job.id))
 
         # 未指定日期且今天失败或无数据时，自动回退到昨天
         need_fallback = not date_str and (
@@ -76,7 +82,13 @@ def crawl(
             else:
                 typer.echo(f"🔄 {target} 无数据，尝试 {fallback} ...")
                 target = fallback
-                result = asyncio.run(crawl_daily(db, target, top_n))
+                job = create_job(
+                    db,
+                    "crawl_daily",
+                    owner="cli_crawl",
+                    payload={"target_date": target, "top_n": top_n},
+                )
+                result = asyncio.run(run_job(db, job.id))
 
         if result["status"] == "success":
             typer.echo(
@@ -110,7 +122,7 @@ def summarize(
     from app.config import settings
     from app.database import SessionLocal, engine
     from app.database import init_db as _init
-    from app.services.summarizer import summarize_batch, summarize_single
+    from app.services.jobs import create_job, run_job
 
     import os
 
@@ -142,11 +154,25 @@ def summarize(
     try:
         if arxiv_id:
             typer.echo(f"🤖 开始总结 {arxiv_id} (mode={pdf_mode}) ...")
-            result = asyncio.run(summarize_single(db, arxiv_id, pdf_mode=pdf_mode))
+            job = create_job(
+                db,
+                "summarize_one",
+                owner="cli_summarize",
+                payload={"arxiv_id": arxiv_id, "pdf_mode": pdf_mode, "force": False},
+            )
         else:
             typer.echo(f"🤖 开始批量总结 pending 论文 (mode={pdf_mode}) ...")
-            result = asyncio.run(summarize_batch(db, pdf_mode=pdf_mode))
+            job = create_job(
+                db,
+                "summarize_batch",
+                owner="cli_summarize",
+                payload={"pdf_mode": pdf_mode},
+            )
 
+        result = asyncio.run(run_job(db, job.id))
+        if result.get("status") == "failed":
+            typer.echo(f"❌ 总结失败：{result.get('error')}", err=True)
+            raise typer.Exit(code=1)
         typer.echo(f"✅ 总结完成：{result}")
     except NotFoundError as exc:
         typer.echo(f"❌ {exc.message}", err=True)
@@ -172,5 +198,39 @@ def init_db():
     typer.echo(f"✅ 数据库已初始化：{settings.db_path}")
 
 
+@cli_app.command("rebuild-derived")
+def rebuild_derived(
+    fts: bool = typer.Option(False, "--fts", help="重建 FTS5 全文索引"),
+    chroma: bool = typer.Option(False, "--chroma", help="重建 ChromaDB 语义索引"),
+):
+    """重建可派生数据索引。"""
+    from app.config import settings
+    from app.database import SessionLocal, engine
+    from app.database import init_db as _init
+    from app.services.jobs import create_job, run_job
+
+    import os
+
+    if not fts and not chroma:
+        fts = True
+
+    os.makedirs(settings.db_path.parent, exist_ok=True)
+    _init(engine)
+
+    db = SessionLocal()
+    try:
+        for job_type in [
+            *(["reindex_fts"] if fts else []),
+            *(["reindex_chroma"] if chroma else []),
+        ]:
+            job = create_job(db, job_type, owner="cli_rebuild_derived", payload={})
+            result = asyncio.run(run_job(db, job.id))
+            typer.echo(f"{job_type}: {result}")
+            if result.get("status") == "failed":
+                raise typer.Exit(code=1)
+    finally:
+        db.close()
+
+
 if __name__ == "__main__":
     cli_app()
diff --git a/app/database.py b/app/database.py
index be456a4..d8cdda0 100644
--- a/app/database.py
+++ b/app/database.py
@@ -76,10 +76,26 @@ def _migrate(engine) -> None:
         "crawl_logs": [
             ("details_json", "TEXT"),
         ],
+        "task_locks": [
+            ("heartbeat_at", "DATETIME"),
+            ("expires_at", "DATETIME"),
+        ],
+        "jobs": [
+            ("heartbeat_at", "DATETIME"),
+        ],
     }
 
     with engine.connect() as conn:
         for table, columns in _MIGRATIONS.items():
+            table_exists = conn.execute(
+                text(
+                    "SELECT name FROM sqlite_master "
+                    "WHERE type IN ('table', 'virtual table') AND name = :name"
+                ),
+                {"name": table},
+            ).fetchone()
+            if not table_exists:
+                continue
             # 获取已有列名
             existing = {
                 row[1] for row in conn.execute(text(f"PRAGMA table_info({table})"))
diff --git a/app/main.py b/app/main.py
index af0564e..396d274 100644
--- a/app/main.py
+++ b/app/main.py
@@ -32,7 +32,14 @@ async def lifespan(app: FastAPI):
     # ── startup ──
     from app.services.scheduler import start_scheduler
     from app.services.embedder import init_chroma
+    from app.services.jobs import recover_stale_jobs
+    from app.database import SessionLocal
 
+    db = SessionLocal()
+    try:
+        recover_stale_jobs(db)
+    finally:
+        db.close()
     start_scheduler()
     init_chroma()
 
diff --git a/app/models.py b/app/models.py
index 2054619..d0673af 100644
--- a/app/models.py
+++ b/app/models.py
@@ -32,6 +32,26 @@ class SummaryState(StrEnum):
     PERMANENT_FAILURE = "permanent_failure"
 
 
+class JobStatus(StrEnum):
+    """后台任务状态枚举 — 对应 jobs.status 列。"""
+
+    QUEUED = "queued"
+    RUNNING = "running"
+    SUCCESS = "success"
+    FAILED = "failed"
+    STALE = "stale"
+    CANCELLED = "cancelled"
+
+
+class JobEventStatus(StrEnum):
+    """任务阶段事件状态枚举 — 对应 job_events.status 列。"""
+
+    STARTED = "started"
+    SUCCESS = "success"
+    FAILED = "failed"
+    INFO = "info"
+
+
 # ── papers ──────────────────────────────────────────────────────────────
 class Paper(Base):
     __tablename__ = "papers"
@@ -194,9 +214,48 @@ class TaskLock(Base):
     status = Column(String, nullable=False)
     owner = Column(String)
     acquired_at = Column(DateTime, nullable=False)
+    heartbeat_at = Column(DateTime)
+    expires_at = Column(DateTime)
     released_at = Column(DateTime)
 
 
+# ── jobs / job_events ──────────────────────────────────────────────────
+class Job(Base):
+    __tablename__ = "jobs"
+
+    id = Column(Integer, primary_key=True, autoincrement=True)
+    type = Column(String, nullable=False, index=True)
+    status = Column(String, nullable=False, default=JobStatus.QUEUED, index=True)
+    owner = Column(String)
+    payload_json = Column(Text)
+    result_json = Column(Text)
+    error = Column(Text)
+    created_at = Column(DateTime, nullable=False)
+    started_at = Column(DateTime)
+    heartbeat_at = Column(DateTime)
+    completed_at = Column(DateTime)
+
+    events = relationship(
+        "JobEvent", back_populates="job", cascade="all, delete-orphan"
+    )
+
+
+class JobEvent(Base):
+    __tablename__ = "job_events"
+
+    id = Column(Integer, primary_key=True, autoincrement=True)
+    job_id = Column(
+        Integer, ForeignKey("jobs.id", ondelete="CASCADE"), nullable=False, index=True
+    )
+    stage = Column(String, nullable=False)
+    status = Column(String, nullable=False)
+    message = Column(Text)
+    payload_json = Column(Text)
+    created_at = Column(DateTime, nullable=False)
+
+    job = relationship("Job", back_populates="events")
+
+
 # ── user data ──────────────────────────────────────────────────────────
 class UserBookmark(Base):
     __tablename__ = "user_bookmarks"
diff --git a/app/routes/admin.py b/app/routes/admin.py
index d1f4873..d69d6dd 100644
--- a/app/routes/admin.py
+++ b/app/routes/admin.py
@@ -4,36 +4,20 @@ from __future__ import annotations
 
 import hashlib
 import hmac
-import json
-import logging
 from datetime import date
 
-from fastapi import APIRouter, Depends, Form, HTTPException, Query, Request
+from fastapi import APIRouter, BackgroundTasks, Depends, Form, HTTPException, Query, Request
 from fastapi.responses import RedirectResponse
 from pydantic import BaseModel, field_validator
-from sqlalchemy import bindparam, func, select, text
 from sqlalchemy.orm import Session
 
 from app.config import settings
 from app.database import get_db
-from app.models import (
-    CrawlLog,
-    DataDeleteJob,
-    Paper,
-    PaperTag,
-    SummaryState,
-    SummaryStatus,
-)
 from app.services import admin as admin_svc
 from app.services.admin import get_admin_stats
-from app.services.cleaner import cleanup_tmp, delete_papers_by_date_range
-from app.services.crawler import refresh_upvotes
-from app.services.pipeline import run_crawl, run_pipeline
-from app.services.scheduler import get_scheduler
-from app.services.summarizer import summarize_batch, summarize_single
-from app.utils import templates, today_str, utc_now
-
-logger = logging.getLogger(__name__)
+from app.services.cleaner import cleanup_tmp
+from app.services.jobs import create_job, enqueue_job
+from app.utils import templates, today_str
 
 router = APIRouter(prefix="/admin", tags=["admin"])
 
@@ -103,18 +87,7 @@ async def admin_dashboard(
 ):
     """管理仪表盘 — 系统状态总览。"""
     stats = get_admin_stats(db)
-
-    # 调度器历史（最近 10 条 task=scheduler 日志）
-    scheduler_history = (
-        db.execute(
-            select(CrawlLog)
-            .where(CrawlLog.task == "scheduler")
-            .order_by(CrawlLog.started_at.desc())
-            .limit(10)
-        )
-        .scalars()
-        .all()
-    )
+    scheduler_history = admin_svc.get_scheduler_history(db)
 
     return templates.TemplateResponse(
         request,
@@ -129,53 +102,43 @@ async def admin_dashboard(
 @router.get("/scheduler-status")
 async def admin_scheduler_status(_admin: None = Depends(verify_admin)):
     """调度器运行状态（JSON）。"""
-    scheduler = get_scheduler()
-    next_run = None
-    upvote_next_run = None
-    if scheduler:
-        for job in scheduler.get_jobs():
-            if job.id == "daily_pipeline":
-                next_run = job.next_run_time
-            elif job.id == "upvote_refresh":
-                upvote_next_run = job.next_run_time
-    return {
-        "enabled": scheduler is not None,
-        "schedule_time": f"{settings.SCHEDULE_HOUR:02d}:{settings.SCHEDULE_MINUTE:02d}",
-        "timezone": settings.APP_TIMEZONE,
-        "next_run": next_run.isoformat() if next_run else None,
-        "upvote_next_run": upvote_next_run.isoformat() if upvote_next_run else None,
-        "upvote_refresh_days": settings.UPVOTE_REFRESH_DAYS,
-    }
+    return admin_svc.get_scheduler_status()
 
 
 @router.post("/trigger-pipeline")
 async def admin_trigger_pipeline(
+    background_tasks: BackgroundTasks,
     _admin: None = Depends(verify_admin),
     db: Session = Depends(get_db),
 ):
     """手动触发一次完整流水线（crawl → summarize → cleanup）。"""
     today = today_str()
-    try:
-        result = await run_pipeline(db, today, owner="admin_trigger")
-    except RuntimeError as exc:
-        raise HTTPException(status_code=409, detail=str(exc))
-
-    if result["status"] == "failed":
-        raise HTTPException(status_code=500, detail=result.get("error"))
-    return {"status": "success", "message": "流水线执行完成"}
+    job = create_job(
+        db,
+        "pipeline_daily",
+        owner="admin_trigger",
+        payload={"target_date": today},
+    )
+    enqueue_job(background_tasks, job.id)
+    return {"status": "queued", "job_id": job.id, "message": "流水线任务已创建"}
 
 
 @router.post("/refresh-upvotes")
 async def admin_refresh_upvotes(
+    background_tasks: BackgroundTasks,
     _admin: None = Depends(verify_admin),
     db: Session = Depends(get_db),
     days: int | None = Query(None, description="刷新最近 N 天，默认使用配置值"),
 ):
     """手动刷新最近 N 天论文的 upvotes。"""
-    result = await refresh_upvotes(db, days=days)
-    if result["status"] == "failed":
-        raise HTTPException(status_code=500, detail=result.get("error"))
-    return result
+    job = create_job(
+        db,
+        "refresh_upvotes",
+        owner="admin_refresh",
+        payload={"days": days},
+    )
+    enqueue_job(background_tasks, job.id)
+    return {"status": "queued", "job_id": job.id}
 
 
 # ── 请求模型 ──────────────────────────────────────────────────────────
@@ -200,18 +163,21 @@ class DeleteRequest(BaseModel):
 
 @router.post("/crawl")
 async def admin_crawl(
+    background_tasks: BackgroundTasks,
     _admin: None = Depends(verify_admin),
     db: Session = Depends(get_db),
     date: str | None = Query(None, description="YYYY-MM-DD，默认今天"),
 ):
     """手动抓取指定日期，默认今天。"""
     target_date = date or today_str()
-    try:
-        return await run_crawl(db, target_date, owner="admin_crawl")
-    except RuntimeError as exc:
-        raise HTTPException(status_code=409, detail=str(exc))
-    except Exception as exc:
-        raise HTTPException(status_code=500, detail=str(exc))
+    job = create_job(
+        db,
+        "crawl_daily",
+        owner="admin_crawl",
+        payload={"target_date": target_date},
+    )
+    enqueue_job(background_tasks, job.id)
+    return {"status": "queued", "job_id": job.id, "target_date": target_date}
 
 
 # ── 总结 ──────────────────────────────────────────────────────────────
@@ -219,23 +185,41 @@ async def admin_crawl(
 
 @router.post("/summarize")
 async def admin_summarize_batch(
+    background_tasks: BackgroundTasks,
     _admin: None = Depends(verify_admin),
     db: Session = Depends(get_db),
 ):
     """批量总结所有 pending 论文。"""
-    return await summarize_batch(db, pdf_mode=settings.SUMMARY_PDF_MODE)
+    job = create_job(
+        db,
+        "summarize_batch",
+        owner="admin_summarize",
+        payload={"pdf_mode": settings.SUMMARY_PDF_MODE},
+    )
+    enqueue_job(background_tasks, job.id)
+    return {"status": "queued", "job_id": job.id}
 
 
 @router.post("/summarize/{arxiv_id}")
 async def admin_summarize_single(
     arxiv_id: str,
+    background_tasks: BackgroundTasks,
     _admin: None = Depends(verify_admin),
     db: Session = Depends(get_db),
 ):
     """总结或重跑单篇论文。"""
-    return await summarize_single(
-        db, arxiv_id, force=True, pdf_mode=settings.SUMMARY_PDF_MODE
+    job = create_job(
+        db,
+        "summarize_one",
+        owner="admin_summarize",
+        payload={
+            "arxiv_id": arxiv_id,
+            "force": True,
+            "pdf_mode": settings.SUMMARY_PDF_MODE,
+        },
     )
+    enqueue_job(background_tasks, job.id)
+    return {"status": "queued", "job_id": job.id, "arxiv_id": arxiv_id}
 
 
 # ── 清理 ──────────────────────────────────────────────────────────────
@@ -243,39 +227,25 @@ async def admin_summarize_single(
 
 @router.post("/cleanup")
 async def admin_cleanup(
+    background_tasks: BackgroundTasks,
     _admin: None = Depends(verify_admin),
     db: Session = Depends(get_db),
 ):
     """清理 data/tmp/ 中超过 24 小时的临时文件。"""
-    now = utc_now()
-    log_entry = CrawlLog(
-        task="cleanup",
-        status="running",
-        started_at=now,
-    )
-    db.add(log_entry)
-    db.commit()
+    job = create_job(db, "cleanup_tmp", owner="admin_cleanup", payload={})
+    enqueue_job(background_tasks, job.id)
+    return {"status": "queued", "job_id": job.id}
 
+
+@router.post("/cleanup-now")
+async def admin_cleanup_now(
+    _admin: None = Depends(verify_admin),
+    db: Session = Depends(get_db),
+):
+    """同步清理临时文件，保留给测试和本地排障使用。"""
     try:
-        result = cleanup_tmp()
-        log_entry.status = "success"
-        log_entry.completed_at = utc_now()
-        log_entry.details_json = json.dumps(
-            {
-                "scanned": result.get("scanned", 0),
-                "removed": result.get("removed", 0),
-            },
-            ensure_ascii=False,
-        )
-        if result.get("errors"):
-            log_entry.error = "; ".join(result["errors"])[:2000]
-        db.commit()
-        return result
+        return admin_svc.run_cleanup_now(db, cleanup_tmp)
     except Exception as exc:
-        log_entry.status = "failed"
-        log_entry.error = str(exc)[:2000]
-        log_entry.completed_at = utc_now()
-        db.commit()
         raise HTTPException(status_code=500, detail=str(exc))
 
 
@@ -285,6 +255,7 @@ async def admin_cleanup(
 @router.post("/delete")
 async def admin_delete(
     body: DeleteRequest,
+    background_tasks: BackgroundTasks,
     _admin: None = Depends(verify_admin),
     db: Session = Depends(get_db),
 ):
@@ -292,13 +263,31 @@ async def admin_delete(
     if body.date_start > body.date_end:
         raise HTTPException(status_code=400, detail="date_start must be <= date_end")
 
-    result = await delete_papers_by_date_range(
+    job = create_job(
         db,
-        body.date_start,
-        body.date_end,
-        include_notes=body.include_notes,
+        "delete_range",
+        owner="admin_delete",
+        payload={
+            "date_start": body.date_start.isoformat(),
+            "date_end": body.date_end.isoformat(),
+            "include_notes": body.include_notes,
+        },
     )
-    return result
+    enqueue_job(background_tasks, job.id)
+    return {"status": "queued", "job_id": job.id}
+
+
+@router.get("/jobs/{job_id}")
+async def admin_job_detail(
+    job_id: int,
+    _admin: None = Depends(verify_admin),
+    db: Session = Depends(get_db),
+):
+    """查询后台任务状态和阶段事件。"""
+    detail = admin_svc.get_job_detail(db, job_id)
+    if not detail:
+        raise HTTPException(status_code=404, detail=f"Job not found: {job_id}")
+    return detail
 
 
 # ── 日志 ──────────────────────────────────────────────────────────────
@@ -313,72 +302,10 @@ async def admin_logs(
     per_page: int = Query(20, ge=1, le=100),
 ):
     """查看任务日志（CrawlLog + DataDeleteJob）+ 总结状态统计。"""
-    crawl_logs = (
-        db.execute(
-            select(CrawlLog)
-            .order_by(CrawlLog.started_at.desc())
-            .limit(per_page)
-            .offset((page - 1) * per_page)
-        )
-        .scalars()
-        .all()
-    )
-
-    delete_jobs = (
-        db.execute(
-            select(DataDeleteJob)
-            .order_by(DataDeleteJob.started_at.desc())
-            .limit(per_page)
-            .offset((page - 1) * per_page)
-        )
-        .scalars()
-        .all()
-    )
-
-    # 总结状态统计概要
-    summary_total = db.scalar(select(func.count(Paper.id))) or 0
-    summary_done = (
-        db.scalar(
-            select(func.count(SummaryStatus.id)).where(
-                SummaryStatus.status == SummaryState.DONE
-            )
-        )
-        or 0
-    )
-    summary_pending = (
-        db.scalar(
-            select(func.count(SummaryStatus.id)).where(
-                SummaryStatus.status.in_(
-                    [SummaryState.PENDING, SummaryState.PROCESSING]
-                )
-            )
-        )
-        or 0
-    )
-    summary_failed = (
-        db.scalar(
-            select(func.count(SummaryStatus.id)).where(
-                SummaryStatus.status.in_(
-                    [SummaryState.FAILED, SummaryState.PERMANENT_FAILURE]
-                )
-            )
-        )
-        or 0
-    )
-
     return templates.TemplateResponse(
         request,
         "admin_logs.html",
-        {
-            "crawl_logs": crawl_logs,
-            "delete_jobs": delete_jobs,
-            "page": page,
-            "per_page": per_page,
-            "summary_total": summary_total,
-            "summary_done": summary_done,
-            "summary_pending": summary_pending,
-            "summary_failed": summary_failed,
-        },
+        admin_svc.get_logs_context(db, page=page, per_page=per_page),
     )
 
 
@@ -395,22 +322,10 @@ async def admin_summary_status(
     per_page: int = Query(20, ge=1, le=100),
 ):
     """总结状态列表（HTMX 片段或 JSON）。"""
-
-    query = (
-        select(Paper, SummaryStatus)
-        .outerjoin(SummaryStatus, SummaryStatus.paper_id == Paper.id)
-        .order_by(Paper.paper_date.desc())
+    results, total = admin_svc.query_summary_statuses(
+        db, status=status, page=page, per_page=per_page
     )
 
-    if status != "all":
-        if status == "none":
-            query = query.where(SummaryStatus.paper_id == None)  # noqa: E711
-        else:
-            query = query.where(SummaryStatus.status == status)
-
-    total = db.scalar(select(func.count()).select_from(query.subquery()))
-    results = db.execute(query.offset((page - 1) * per_page).limit(per_page)).all()
-
     # 判断是否 HTMX 请求
     is_htmx = request.headers.get("HX-Request") == "true"
 
@@ -421,27 +336,16 @@ async def admin_summary_status(
             "partials/summary_list.html",
             {
                 "results": results,
-                "total": total or 0,
+                "total": total,
                 "page": page,
                 "per_page": per_page,
                 "current_status": status,
             },
         )
 
-    # 非 HTMX 返回 JSON
-    items = []
-    for paper, ss in results:
-        item = {
-            "arxiv_id": paper.arxiv_id,
-            "title": paper.title_zh or paper.title_en,
-            "paper_date": str(paper.paper_date),
-            "summary_status": ss.status if ss else "none",
-            "retry_count": ss.retry_count if ss else 0,
-            "error_type": ss.error_type if ss else None,
-            "error": ss.error if ss else None,
-        }
-        items.append(item)
-    return {"items": items, "total": total or 0, "page": page, "per_page": per_page}
+    return admin_svc.serialize_summary_statuses(
+        results, total=total, page=page, per_page=per_page
+    )
 
 
 @router.post("/summary-retry-failed")
@@ -450,39 +354,14 @@ async def admin_summary_retry_failed(
     db: Session = Depends(get_db),
 ):
     """重试所有失败状态的总结任务。"""
-    failed_ids = (
-        db.execute(
-            select(Paper.arxiv_id)
-            .join(SummaryStatus, SummaryStatus.paper_id == Paper.id)
-            .where(
-                SummaryStatus.status.in_(
-                    [SummaryState.FAILED, SummaryState.PERMANENT_FAILURE]
-                )
-            )
-        )
-        .scalars()
-        .all()
-    )
-
-    if not failed_ids:
+    count = admin_svc.retry_failed_summaries(db)
+    if not count:
         return {"status": "success", "message": "没有失败的任务需要重试", "count": 0}
 
-    # 重置失败任务的状态为 pending
-    db.execute(
-        SummaryStatus.__table__.update()
-        .where(
-            SummaryStatus.status.in_(
-                [SummaryState.FAILED, SummaryState.PERMANENT_FAILURE]
-            )
-        )
-        .values(status=SummaryState.PENDING, error=None, error_type=None)
-    )
-    db.commit()
-
     return {
         "status": "success",
-        "message": f"已重置 {len(failed_ids)} 个失败任务为待总结状态",
-        "count": len(failed_ids),
+        "message": f"已重置 {count} 个失败任务为待总结状态",
+        "count": count,
     }
 
 
@@ -545,23 +424,8 @@ async def admin_paper_delete(
     db: Session = Depends(get_db),
 ):
     """删除单篇论文。"""
-    paper = db.scalar(select(Paper).where(Paper.arxiv_id == arxiv_id))
-    if not paper:
+    if not admin_svc.delete_paper_by_arxiv(db, arxiv_id):
         raise HTTPException(status_code=404, detail=f"Paper not found: {arxiv_id}")
-
-    # 删除相关数据（ORM cascade 自动处理关联表）
-    db.delete(paper)
-    db.commit()
-
-    # 清理 FTS 索引
-    try:
-        db.execute(
-            text("DELETE FROM papers_fts WHERE arxiv_id = :aid"), {"aid": arxiv_id}
-        )
-        db.commit()
-    except Exception:
-        logger.warning("Failed to clean FTS index for %s", arxiv_id, exc_info=True)
-
     return {"status": "success", "message": f"已删除 {arxiv_id}"}
 
 
@@ -588,28 +452,7 @@ async def admin_papers_batch_action(
         raise HTTPException(status_code=400, detail="arxiv_ids 不能为空")
 
     if body.action == "delete":
-        papers = (
-            db.execute(select(Paper).where(Paper.arxiv_id.in_(body.arxiv_ids)))
-            .scalars()
-            .all()
-        )
-
-        count = 0
-        for paper in papers:
-            db.delete(paper)
-            count += 1
-        db.commit()
-
-        # 清理 FTS 索引
-        try:
-            stmt = text("DELETE FROM papers_fts WHERE arxiv_id IN :ids").bindparams(
-                bindparam("ids", expanding=True)
-            )
-            db.execute(stmt, {"ids": body.arxiv_ids})
-            db.commit()
-        except Exception:
-            logger.warning("Failed to clean FTS index for batch delete", exc_info=True)
-
+        count = admin_svc.delete_papers_by_arxiv_ids(db, body.arxiv_ids)
         return {
             "status": "success",
             "message": f"已删除 {count} 篇论文",
@@ -617,24 +460,10 @@ async def admin_papers_batch_action(
         }
 
     elif body.action == "summarize":
-        # 将选中论文的总结状态重置为 pending
-        paper_ids = (
-            db.execute(select(Paper.id).where(Paper.arxiv_id.in_(body.arxiv_ids)))
-            .scalars()
-            .all()
-        )
-
-        if paper_ids:
-            # 删除旧的 status 记录让其重新进入 pipeline
-            db.execute(
-                SummaryStatus.__table__.delete().where(
-                    SummaryStatus.paper_id.in_(paper_ids)
-                )
-            )
-            db.commit()
+        count = admin_svc.reset_summaries_pending(db, body.arxiv_ids)
 
         return {
             "status": "success",
-            "message": f"已将 {len(paper_ids)} 篇论文重置为待总结",
-            "count": len(paper_ids),
+            "message": f"已将 {count} 篇论文重置为待总结",
+            "count": count,
         }
diff --git a/app/services/admin.py b/app/services/admin.py
index 20d8e09..468d753 100644
--- a/app/services/admin.py
+++ b/app/services/admin.py
@@ -1,17 +1,30 @@
-"""管理后台服务 — 统计聚合、系统状态。"""
+"""管理后台服务 — 统计聚合、系统状态、管理操作。"""
 
 from __future__ import annotations
 
+import json
 from datetime import date
 from pathlib import Path
+from typing import Callable
 
 from sqlalchemy import func, select, text
 from sqlalchemy.orm import Session
 
 from app.config import settings
-from app.models import CrawlLog, Paper, PaperTag, SummaryState, SummaryStatus, TaskLock
+from app.models import (
+    CrawlLog,
+    DataDeleteJob,
+    Job,
+    JobEvent,
+    Paper,
+    PaperTag,
+    SummaryState,
+    SummaryStatus,
+    TaskLock,
+)
+from app.services.derived import delete_paper_indexes
 from app.services.scheduler import get_scheduler
-from app.utils import PAPERS_DIR, TMP_DIR
+from app.utils import PAPERS_DIR, TMP_DIR, utc_now
 
 # admin_papers 排序映射
 SORT_MAP = {
@@ -190,3 +203,311 @@ def query_papers(
             statuses[paper_id_to_arxiv.get(pid, "")] = st
 
     return papers, total or 0, statuses
+
+
+def get_scheduler_history(db: Session, limit: int = 10) -> list[CrawlLog]:
+    """最近的调度器运行日志。"""
+    return (
+        db.execute(
+            select(CrawlLog)
+            .where(CrawlLog.task == "scheduler")
+            .order_by(CrawlLog.started_at.desc())
+            .limit(limit)
+        )
+        .scalars()
+        .all()
+    )
+
+
+def get_scheduler_status() -> dict:
+    """调度器运行状态。"""
+    scheduler = get_scheduler()
+    next_run = None
+    upvote_next_run = None
+    if scheduler:
+        for job in scheduler.get_jobs():
+            if job.id == "daily_pipeline":
+                next_run = job.next_run_time
+            elif job.id == "upvote_refresh":
+                upvote_next_run = job.next_run_time
+    return {
+        "enabled": scheduler is not None,
+        "schedule_time": f"{settings.SCHEDULE_HOUR:02d}:{settings.SCHEDULE_MINUTE:02d}",
+        "timezone": settings.APP_TIMEZONE,
+        "next_run": next_run.isoformat() if next_run else None,
+        "upvote_next_run": upvote_next_run.isoformat() if upvote_next_run else None,
+        "upvote_refresh_days": settings.UPVOTE_REFRESH_DAYS,
+    }
+
+
+def run_cleanup_now(db: Session, cleanup_func: Callable[[], dict]) -> dict:
+    """同步执行临时目录清理，并写入 CrawlLog。"""
+    log_entry = CrawlLog(task="cleanup", status="running", started_at=utc_now())
+    db.add(log_entry)
+    db.commit()
+
+    try:
+        result = cleanup_func()
+        log_entry.status = "success"
+        log_entry.completed_at = utc_now()
+        log_entry.details_json = json.dumps(
+            {
+                "scanned": result.get("scanned", 0),
+                "removed": result.get("removed", 0),
+            },
+            ensure_ascii=False,
+        )
+        if result.get("errors"):
+            log_entry.error = "; ".join(result["errors"])[:2000]
+        db.commit()
+        return result
+    except Exception as exc:
+        log_entry.status = "failed"
+        log_entry.error = str(exc)[:2000]
+        log_entry.completed_at = utc_now()
+        db.commit()
+        raise
+
+
+def get_job_detail(db: Session, job_id: int) -> dict | None:
+    """后台任务详情和阶段事件，返回可 JSON 序列化 dict。"""
+    job = db.get(Job, job_id)
+    if not job:
+        return None
+    events = (
+        db.execute(
+            select(JobEvent)
+            .where(JobEvent.job_id == job_id)
+            .order_by(JobEvent.created_at.asc())
+        )
+        .scalars()
+        .all()
+    )
+    return {
+        "id": job.id,
+        "type": job.type,
+        "status": job.status,
+        "owner": job.owner,
+        "payload": json.loads(job.payload_json or "{}"),
+        "result": json.loads(job.result_json or "{}") if job.result_json else None,
+        "error": job.error,
+        "created_at": job.created_at.isoformat(),
+        "started_at": job.started_at.isoformat() if job.started_at else None,
+        "completed_at": job.completed_at.isoformat() if job.completed_at else None,
+        "events": [
+            {
+                "stage": event.stage,
+                "status": event.status,
+                "message": event.message,
+                "payload": json.loads(event.payload_json or "{}")
+                if event.payload_json
+                else None,
+                "created_at": event.created_at.isoformat(),
+            }
+            for event in events
+        ],
+    }
+
+
+def get_logs_context(db: Session, *, page: int, per_page: int) -> dict:
+    """管理日志页上下文。"""
+    crawl_logs = (
+        db.execute(
+            select(CrawlLog)
+            .order_by(CrawlLog.started_at.desc())
+            .limit(per_page)
+            .offset((page - 1) * per_page)
+        )
+        .scalars()
+        .all()
+    )
+    delete_jobs = (
+        db.execute(
+            select(DataDeleteJob)
+            .order_by(DataDeleteJob.started_at.desc())
+            .limit(per_page)
+            .offset((page - 1) * per_page)
+        )
+        .scalars()
+        .all()
+    )
+
+    summary_total = db.scalar(select(func.count(Paper.id))) or 0
+    summary_done = (
+        db.scalar(
+            select(func.count(SummaryStatus.id)).where(
+                SummaryStatus.status == SummaryState.DONE
+            )
+        )
+        or 0
+    )
+    summary_pending = (
+        db.scalar(
+            select(func.count(SummaryStatus.id)).where(
+                SummaryStatus.status.in_(
+                    [SummaryState.PENDING, SummaryState.PROCESSING]
+                )
+            )
+        )
+        or 0
+    )
+    summary_failed = (
+        db.scalar(
+            select(func.count(SummaryStatus.id)).where(
+                SummaryStatus.status.in_(
+                    [SummaryState.FAILED, SummaryState.PERMANENT_FAILURE]
+                )
+            )
+        )
+        or 0
+    )
+    return {
+        "crawl_logs": crawl_logs,
+        "delete_jobs": delete_jobs,
+        "page": page,
+        "per_page": per_page,
+        "summary_total": summary_total,
+        "summary_done": summary_done,
+        "summary_pending": summary_pending,
+        "summary_failed": summary_failed,
+    }
+
+
+def query_summary_statuses(
+    db: Session,
+    *,
+    status: str,
+    page: int,
+    per_page: int,
+) -> tuple[list[tuple[Paper, SummaryStatus | None]], int]:
+    """总结状态列表查询。"""
+    query = (
+        select(Paper, SummaryStatus)
+        .outerjoin(SummaryStatus, SummaryStatus.paper_id == Paper.id)
+        .order_by(Paper.paper_date.desc())
+    )
+    if status != "all":
+        if status == "none":
+            query = query.where(SummaryStatus.paper_id == None)  # noqa: E711
+        else:
+            query = query.where(SummaryStatus.status == status)
+
+    total = db.scalar(select(func.count()).select_from(query.subquery())) or 0
+    results = db.execute(query.offset((page - 1) * per_page).limit(per_page)).all()
+    return results, total
+
+
+def serialize_summary_statuses(
+    results: list[tuple[Paper, SummaryStatus | None]],
+    *,
+    total: int,
+    page: int,
+    per_page: int,
+) -> dict:
+    """总结状态列表 JSON 响应。"""
+    items = []
+    for paper, ss in results:
+        items.append(
+            {
+                "arxiv_id": paper.arxiv_id,
+                "title": paper.title_zh or paper.title_en,
+                "paper_date": str(paper.paper_date),
+                "summary_status": ss.status if ss else "none",
+                "retry_count": ss.retry_count if ss else 0,
+                "error_type": ss.error_type if ss else None,
+                "error": ss.error if ss else None,
+            }
+        )
+    return {"items": items, "total": total, "page": page, "per_page": per_page}
+
+
+def retry_failed_summaries(db: Session) -> int:
+    """将失败/永久失败的总结任务重置为 pending。"""
+    failed_ids = (
+        db.execute(
+            select(Paper.arxiv_id)
+            .join(SummaryStatus, SummaryStatus.paper_id == Paper.id)
+            .where(
+                SummaryStatus.status.in_(
+                    [SummaryState.FAILED, SummaryState.PERMANENT_FAILURE]
+                )
+            )
+        )
+        .scalars()
+        .all()
+    )
+    if not failed_ids:
+        return 0
+
+    db.execute(
+        SummaryStatus.__table__.update()
+        .where(
+            SummaryStatus.status.in_(
+                [SummaryState.FAILED, SummaryState.PERMANENT_FAILURE]
+            )
+        )
+        .values(status=SummaryState.PENDING, error=None, error_type=None)
+    )
+    db.commit()
+    return len(failed_ids)
+
+
+def delete_paper_by_arxiv(db: Session, arxiv_id: str) -> bool:
+    """删除单篇论文和派生索引。"""
+    paper = db.scalar(select(Paper).where(Paper.arxiv_id == arxiv_id))
+    if not paper:
+        return False
+
+    paper_id = paper.id
+    db.delete(paper)
+    db.commit()
+    delete_paper_indexes(db, paper_id=paper_id, arxiv_id=arxiv_id)
+    db.commit()
+    return True
+
+
+def delete_papers_by_arxiv_ids(db: Session, arxiv_ids: list[str]) -> int:
+    """批量删除论文和派生索引。"""
+    papers = (
+        db.execute(select(Paper).where(Paper.arxiv_id.in_(arxiv_ids))).scalars().all()
+    )
+    deleted = [(paper.id, paper.arxiv_id) for paper in papers]
+    for paper in papers:
+        db.delete(paper)
+    db.commit()
+
+    for paper_id, arxiv_id in deleted:
+        delete_paper_indexes(db, paper_id=paper_id, arxiv_id=arxiv_id)
+    db.commit()
+    return len(deleted)
+
+
+def reset_summaries_pending(db: Session, arxiv_ids: list[str]) -> int:
+    """将指定论文的总结状态重置为 pending，没有状态则创建。"""
+    paper_ids = (
+        db.execute(select(Paper.id).where(Paper.arxiv_id.in_(arxiv_ids)))
+        .scalars()
+        .all()
+    )
+    if not paper_ids:
+        return 0
+
+    existing_statuses = (
+        db.execute(select(SummaryStatus).where(SummaryStatus.paper_id.in_(paper_ids)))
+        .scalars()
+        .all()
+    )
+    existing_ids = {status.paper_id for status in existing_statuses}
+    for status in existing_statuses:
+        status.status = SummaryState.PENDING
+        status.quality = None
+        status.error = None
+        status.error_type = None
+        status.raw_output_saved = False
+        status.started_at = None
+        status.completed_at = None
+    for paper_id in paper_ids:
+        if paper_id not in existing_ids:
+            db.add(SummaryStatus(paper_id=paper_id, status=SummaryState.PENDING))
+    db.commit()
+    return len(paper_ids)
diff --git a/app/services/crawler.py b/app/services/crawler.py
index 5942373..ca8f23b 100644
--- a/app/services/crawler.py
+++ b/app/services/crawler.py
@@ -4,7 +4,7 @@ import logging
 from datetime import date as date_type, datetime, timezone
 
 import httpx
-from sqlalchemy import select, text
+from sqlalchemy import select
 from sqlalchemy.orm import Session
 
 from app.config import settings
@@ -16,6 +16,7 @@ from app.models import (
     SummaryState,
     SummaryStatus,
 )
+from app.services.derived import reindex_paper_fts
 from app.utils import make_http_client, recent_date_strs, utc_now
 
 logger = logging.getLogger(__name__)
@@ -143,21 +144,7 @@ def upsert_papers(db: Session, papers_raw: list[dict], paper_date: str) -> list[
 
             db.add(SummaryStatus(paper_id=paper.id, status=SummaryState.PENDING))
 
-            authors_text = ", ".join(meta["authors"])
-            tags_text = ", ".join(meta["tags"])
-            db.execute(
-                text(
-                    "INSERT INTO papers_fts(rowid, title_en, abstract, authors, tags) "
-                    "VALUES (:id, :title, :abstract, :authors, :tags)"
-                ),
-                {
-                    "id": paper.id,
-                    "title": meta["title_en"],
-                    "abstract": meta["abstract"] or "",
-                    "authors": authors_text,
-                    "tags": tags_text,
-                },
-            )
+            reindex_paper_fts(db, paper)
 
             new_papers.append(paper)
             logger.debug("Inserted new paper: %s", arxiv_id)
diff --git a/app/services/derived.py b/app/services/derived.py
new file mode 100644
index 0000000..1eece19
--- /dev/null
+++ b/app/services/derived.py
@@ -0,0 +1,140 @@
+"""派生数据维护 — FTS5 / ChromaDB 等可重建索引。"""
+
+from __future__ import annotations
+
+import logging
+
+from sqlalchemy import select, text
+from sqlalchemy.orm import Session
+
+from app.models import Paper
+
+logger = logging.getLogger(__name__)
+
+
+def _summary_text(paper: Paper) -> str:
+    summary = paper.summary
+    if not summary:
+        return ""
+    parts = [
+        summary.one_line,
+        summary.motivation_problem,
+        summary.motivation_goal,
+        summary.method_overview,
+        summary.method_key_idea,
+        summary.results_main_json,
+    ]
+    return " ".join(p for p in parts if p)
+
+
+def delete_fts_paper(db: Session, paper_id: int) -> None:
+    """删除单篇论文的 FTS5 行。FTS5 以 papers.id 作为 rowid。"""
+    db.execute(
+        text("DELETE FROM papers_fts WHERE rowid = :paper_id"),
+        {"paper_id": paper_id},
+    )
+
+
+def delete_paper_indexes(db: Session, *, paper_id: int, arxiv_id: str) -> None:
+    """删除单篇论文的所有派生索引。失败项记录日志但不阻断主删除。"""
+    try:
+        delete_fts_paper(db, paper_id)
+    except Exception:
+        logger.warning("Failed to clean FTS index for %s", arxiv_id, exc_info=True)
+
+    try:
+        from app.services.embedder import delete_paper
+
+        delete_paper(arxiv_id)
+    except Exception:
+        logger.warning("Failed to clean ChromaDB index for %s", arxiv_id, exc_info=True)
+
+
+def reindex_paper_fts(db: Session, paper: Paper) -> None:
+    """按 DB 权威数据重建单篇论文的 FTS5 派生索引。"""
+    authors_text = ", ".join(
+        a.name for a in sorted(paper.authors, key=lambda a: a.position or 0)
+    )
+    tags_text = ", ".join(t.tag for t in paper.tags)
+
+    delete_fts_paper(db, paper.id)
+    db.execute(
+        text(
+            """
+            INSERT INTO papers_fts(
+                rowid, title_en, title_zh, abstract, authors, tags, summary_text
+            )
+            VALUES (
+                :id, :title_en, :title_zh, :abstract, :authors, :tags, :summary_text
+            )
+            """
+        ),
+        {
+            "id": paper.id,
+            "title_en": paper.title_en or "",
+            "title_zh": paper.title_zh or "",
+            "abstract": paper.abstract or "",
+            "authors": authors_text,
+            "tags": tags_text,
+            "summary_text": _summary_text(paper),
+        },
+    )
+
+
+def reindex_fts(db: Session, paper_ids: list[int] | None = None) -> dict:
+    """全量或局部重建 FTS5 索引。"""
+    query = select(Paper)
+    if paper_ids:
+        query = query.where(Paper.id.in_(paper_ids))
+    papers = db.execute(query).scalars().all()
+
+    if paper_ids is None:
+        db.execute(text("DELETE FROM papers_fts"))
+
+    count = 0
+    for paper in papers:
+        reindex_paper_fts(db, paper)
+        count += 1
+    db.commit()
+    logger.info("FTS reindexed: %d papers", count)
+    return {"status": "success", "indexed": count}
+
+
+def reindex_chroma(db: Session) -> dict:
+    """按 DB 权威数据重建 ChromaDB 语义索引。"""
+    from app.services.embedder import index_paper
+
+    papers = db.execute(select(Paper).where(Paper.summary.has())).scalars().all()
+    indexed = 0
+    errors: list[str] = []
+    for paper in papers:
+        try:
+            texts_dict = {
+                "arxiv_id": paper.arxiv_id,
+                "title_zh": paper.title_zh or "",
+                "title_en": paper.title_en or "",
+                "tags": " ".join(t.tag for t in paper.tags),
+                "one_line": paper.summary.one_line if paper.summary else "",
+                "motivation_problem": (
+                    paper.summary.motivation_problem if paper.summary else ""
+                ),
+                "method_key_idea": (
+                    paper.summary.method_key_idea if paper.summary else ""
+                ),
+                "paper_date": paper.paper_date.isoformat() if paper.paper_date else "",
+            }
+            index_paper(paper.arxiv_id, texts_dict)
+            indexed += 1
+        except Exception as exc:
+            errors.append(f"{paper.arxiv_id}: {exc}")
+            logger.warning(
+                "Failed to reindex ChromaDB for %s",
+                paper.arxiv_id,
+                exc_info=True,
+            )
+
+    return {
+        "status": "success" if not errors else "partial",
+        "indexed": indexed,
+        "errors": errors or None,
+    }
diff --git a/app/services/jobs.py b/app/services/jobs.py
new file mode 100644
index 0000000..e4ac2e9
--- /dev/null
+++ b/app/services/jobs.py
@@ -0,0 +1,244 @@
+"""统一后台任务系统 — 创建、运行、事件记录、失败恢复。"""
+
+from __future__ import annotations
+
+import json
+import logging
+from datetime import date, timedelta
+from typing import Any
+
+from fastapi import BackgroundTasks
+from sqlalchemy import or_, select
+from sqlalchemy.orm import Session
+
+from app.config import settings
+from app.database import SessionLocal
+from app.models import Job, JobEvent, JobEventStatus, JobStatus, TaskLock
+from app.utils import truncate_error, utc_now
+
+logger = logging.getLogger(__name__)
+
+STALE_JOB_AFTER = timedelta(hours=6)
+
+
+def _dumps(value: Any) -> str:
+    return json.dumps(value, ensure_ascii=False, default=str)
+
+
+def _loads(value: str | None) -> dict:
+    if not value:
+        return {}
+    try:
+        data = json.loads(value)
+        return data if isinstance(data, dict) else {}
+    except json.JSONDecodeError:
+        return {}
+
+
+def create_job(
+    db: Session,
+    job_type: str,
+    *,
+    owner: str,
+    payload: dict | None = None,
+) -> Job:
+    """创建后台任务主记录。"""
+    job = Job(
+        type=job_type,
+        status=JobStatus.QUEUED,
+        owner=owner,
+        payload_json=_dumps(payload or {}),
+        created_at=utc_now(),
+    )
+    db.add(job)
+    db.commit()
+    db.refresh(job)
+    add_job_event(
+        db,
+        job,
+        stage="created",
+        status=JobEventStatus.INFO,
+        message=f"Job queued: {job_type}",
+        payload=payload or {},
+    )
+    return job
+
+
+def add_job_event(
+    db: Session,
+    job: Job,
+    *,
+    stage: str,
+    status: str,
+    message: str | None = None,
+    payload: dict | None = None,
+) -> None:
+    """追加一条任务阶段事件。"""
+    db.add(
+        JobEvent(
+            job_id=job.id,
+            stage=stage,
+            status=str(status),
+            message=message,
+            payload_json=_dumps(payload) if payload is not None else None,
+            created_at=utc_now(),
+        )
+    )
+    job.heartbeat_at = utc_now()
+    db.commit()
+
+
+def enqueue_job(background_tasks: BackgroundTasks, job_id: int) -> None:
+    """把任务提交给 FastAPI BackgroundTasks。"""
+    background_tasks.add_task(run_job_by_id, job_id)
+
+
+async def run_job_by_id(job_id: int) -> None:
+    """使用独立 DB session 运行一个已创建的 job。"""
+    db = SessionLocal()
+    try:
+        await run_job(db, job_id)
+    finally:
+        db.close()
+
+
+async def run_job(db: Session, job_id: int) -> dict:
+    """运行 job，并把状态/result/error 写回 jobs/job_events。"""
+    job = db.get(Job, job_id)
+    if not job:
+        raise ValueError(f"Job not found: {job_id}")
+    if job.status == JobStatus.RUNNING:
+        raise RuntimeError(f"Job already running: {job_id}")
+
+    payload = _loads(job.payload_json)
+    job.status = JobStatus.RUNNING
+    job.started_at = utc_now()
+    job.heartbeat_at = job.started_at
+    db.commit()
+    add_job_event(db, job, stage="run", status=JobEventStatus.STARTED)
+
+    try:
+        result = await _dispatch_job(db, job, payload)
+    except Exception as exc:
+        logger.exception("Job failed: id=%s type=%s", job.id, job.type)
+        error = truncate_error(exc, limit=4000)
+        job.status = JobStatus.FAILED
+        job.error = error
+        job.completed_at = utc_now()
+        db.commit()
+        add_job_event(db, job, stage="run", status=JobEventStatus.FAILED, message=error)
+        return {"status": "failed", "error": error}
+
+    job.status = JobStatus.SUCCESS
+    job.result_json = _dumps(result)
+    job.completed_at = utc_now()
+    job.error = None
+    db.commit()
+    add_job_event(
+        db,
+        job,
+        stage="run",
+        status=JobEventStatus.SUCCESS,
+        payload=result if isinstance(result, dict) else {"result": result},
+    )
+    return result if isinstance(result, dict) else {"status": "success", "result": result}
+
+
+async def _dispatch_job(db: Session, job: Job, payload: dict) -> dict:
+    from app.services.cleaner import cleanup_tmp, delete_papers_by_date_range
+    from app.services.crawler import refresh_upvotes
+    from app.services.derived import reindex_chroma, reindex_fts
+    from app.services.pipeline import run_crawl, run_pipeline
+    from app.services.summarizer import summarize_batch, summarize_single
+
+    if job.type == "crawl_daily":
+        return await run_crawl(
+            db,
+            payload["target_date"],
+            owner=job.owner or f"job:{job.id}",
+            top_n=payload.get("top_n"),
+        )
+    if job.type == "pipeline_daily":
+        return await run_pipeline(
+            db,
+            payload["target_date"],
+            owner=job.owner or f"job:{job.id}",
+        )
+    if job.type == "summarize_batch":
+        return await summarize_batch(
+            db,
+            pdf_mode=payload.get("pdf_mode", settings.SUMMARY_PDF_MODE),
+        )
+    if job.type == "summarize_one":
+        return await summarize_single(
+            db,
+            payload["arxiv_id"],
+            force=payload.get("force", True),
+            pdf_mode=payload.get("pdf_mode", settings.SUMMARY_PDF_MODE),
+        )
+    if job.type == "refresh_upvotes":
+        return await refresh_upvotes(db, days=payload.get("days"))
+    if job.type == "delete_range":
+        return await delete_papers_by_date_range(
+            db,
+            date.fromisoformat(payload["date_start"]),
+            date.fromisoformat(payload["date_end"]),
+            include_notes=payload.get("include_notes", True),
+        )
+    if job.type == "cleanup_tmp":
+        return cleanup_tmp()
+    if job.type == "reindex_fts":
+        return reindex_fts(db)
+    if job.type == "reindex_chroma":
+        return reindex_chroma(db)
+
+    raise ValueError(f"Unsupported job type: {job.type}")
+
+
+def recover_stale_jobs(db: Session) -> int:
+    """启动时将过期 running job/lock 标记为 stale，避免永久卡住。"""
+    now = utc_now()
+    cutoff = now - STALE_JOB_AFTER
+    stale_jobs = (
+        db.execute(
+            select(Job).where(
+                Job.status == JobStatus.RUNNING,
+                or_(Job.heartbeat_at == None, Job.heartbeat_at < cutoff),  # noqa: E711
+            )
+        )
+        .scalars()
+        .all()
+    )
+    for job in stale_jobs:
+        job.status = JobStatus.STALE
+        job.error = "Marked stale after process restart or missed heartbeat"
+        job.completed_at = now
+        db.add(
+            JobEvent(
+                job_id=job.id,
+                stage="recovery",
+                status=JobEventStatus.FAILED,
+                message=job.error,
+                created_at=now,
+            )
+        )
+
+    stale_locks = (
+        db.execute(
+            select(TaskLock).where(
+                TaskLock.status == "running",
+                TaskLock.acquired_at < cutoff,
+            )
+        )
+        .scalars()
+        .all()
+    )
+    for lock in stale_locks:
+        lock.status = "stale"
+        lock.released_at = now
+
+    db.commit()
+    recovered = len(stale_jobs) + len(stale_locks)
+    if recovered:
+        logger.warning("Recovered stale runtime records: %d", recovered)
+    return recovered
diff --git a/app/services/pipeline.py b/app/services/pipeline.py
index e731bf5..c7bee1b 100644
--- a/app/services/pipeline.py
+++ b/app/services/pipeline.py
@@ -7,6 +7,7 @@ from __future__ import annotations
 
 import logging
 from datetime import date as date_type
+from datetime import timedelta
 
 from sqlalchemy.exc import IntegrityError
 from sqlalchemy.orm import Session
@@ -32,6 +33,8 @@ def acquire_lock(db: Session, task: str, lock_key: str, owner: str) -> TaskLock:
         status="running",
         owner=owner,
         acquired_at=utc_now(),
+        heartbeat_at=utc_now(),
+        expires_at=utc_now() + timedelta(hours=6),
     )
     try:
         db.add(lock)
@@ -42,7 +45,12 @@ def acquire_lock(db: Session, task: str, lock_key: str, owner: str) -> TaskLock:
     return lock
 
 
-async def run_crawl(db: Session, target_date: str, owner: str = "admin_crawl") -> dict:
+async def run_crawl(
+    db: Session,
+    target_date: str,
+    owner: str = "admin_crawl",
+    top_n: int | None = None,
+) -> dict:
     """执行单次抓取（带防重入锁）。
 
     Args:
@@ -55,7 +63,7 @@ async def run_crawl(db: Session, target_date: str, owner: str = "admin_crawl") -
     """
     lock = acquire_lock(db, "crawl", target_date, owner)
     try:
-        return await crawl_daily(db, target_date)
+        return await crawl_daily(db, target_date, top_n=top_n)
     finally:
         release_lock(db, lock)
 
@@ -83,6 +91,8 @@ async def run_pipeline(db: Session, target_date: str, owner: str) -> dict:
         status="running",
         owner=owner,
         acquired_at=now,
+        heartbeat_at=now,
+        expires_at=now + timedelta(hours=6),
     )
     try:
         db.add(lock)
diff --git a/app/services/scheduler.py b/app/services/scheduler.py
index 698056d..bae0e0f 100644
--- a/app/services/scheduler.py
+++ b/app/services/scheduler.py
@@ -11,8 +11,7 @@ from zoneinfo import ZoneInfo
 
 from app.config import settings
 from app.database import SessionLocal
-from app.services.pipeline import run_pipeline
-from app.services.crawler import refresh_upvotes
+from app.services.jobs import create_job, run_job
 from app.utils import today_str
 
 logger = logging.getLogger(__name__)
@@ -112,7 +111,13 @@ async def _daily_pipeline() -> None:
 
     db: Session = SessionLocal()
     try:
-        await run_pipeline(db, today, owner="daily_pipeline")
+        job = create_job(
+            db,
+            "pipeline_daily",
+            owner="daily_pipeline",
+            payload={"target_date": today},
+        )
+        await run_job(db, job.id)
     except RuntimeError:
         logger.warning("Daily pipeline already running for %s, skipping", today)
     except Exception:
@@ -125,7 +130,8 @@ async def _upvote_refresh() -> None:
     """刷新最近 N 天论文的 upvotes。"""
     db: Session = SessionLocal()
     try:
-        result = await refresh_upvotes(db)
+        job = create_job(db, "refresh_upvotes", owner="upvote_refresh", payload={})
+        result = await run_job(db, job.id)
         logger.info(
             "Upvote refresh completed: status=%s updated=%d",
             result.get("status"),
diff --git a/app/services/summary_persister.py b/app/services/summary_persister.py
index ef90b4e..1dd99d6 100644
--- a/app/services/summary_persister.py
+++ b/app/services/summary_persister.py
@@ -3,8 +3,6 @@
 from __future__ import annotations
 
 import logging
-
-from sqlalchemy import text
 from sqlalchemy.orm import Session
 
 from app.models import (
@@ -13,6 +11,7 @@ from app.models import (
     PaperTag,
     SummaryState,
 )
+from app.services.derived import reindex_paper_fts
 from app.services.pdf_downloader import paper_dir
 from app.services.schemas import (
     SummarySchema,
@@ -75,19 +74,9 @@ def _update_summary_in_db(
             db.add(PaperTag(paper_id=paper.id, tag=tag_name, source="ai"))
             existing_tag_names.add(tag_name)
 
-    # 4. FTS5 更新
-    summary_text = _build_fts_summary_text(schema)
-    db.execute(
-        text(
-            "UPDATE papers_fts SET title_zh=:title_zh, summary_text=:summary_text "
-            "WHERE rowid=:paper_id"
-        ),
-        {
-            "title_zh": schema.title_zh,
-            "summary_text": summary_text,
-            "paper_id": paper.id,
-        },
-    )
+    # 4. FTS5 派生索引
+    db.flush()
+    reindex_paper_fts(db, paper)
 
     db.commit()
     logger.info("DB updated: paper=%s quality=%s", paper.arxiv_id, quality)
diff --git a/tests/test_admin.py b/tests/test_admin.py
index 4d16de3..999c3f3 100644
--- a/tests/test_admin.py
+++ b/tests/test_admin.py
@@ -3,14 +3,17 @@
 from __future__ import annotations
 
 import logging
-from unittest.mock import AsyncMock, patch
+from unittest.mock import patch
 
 import pytest
-from sqlalchemy import select
+from sqlalchemy import select, text
 
 from app.config import settings
 from app.models import (
     CrawlLog,
+    Job,
+    SummaryState,
+    SummaryStatus,
     TaskLock,
 )
 from app.utils import utc_now
@@ -64,47 +67,13 @@ class TestAdminAuth:
         resp = auth_client.get("/admin/logs", follow_redirects=False)
         assert resp.status_code == 303
 
-    def test_correct_session_accepted(self, auth_client):
-        """已登录 session 应被接受（crawl 可能会失败但不是 303）。"""
-        with patch(
-            "app.routes.admin.run_crawl", new_callable=AsyncMock
-        ) as mock_crawl:
-            mock_crawl.return_value = {"found": 0, "new": 0, "status": "success"}
-            resp = auth_client.post("/admin/crawl")
-            assert resp.status_code != 303
-
-    # ── summarize route auth ────────────────────────────────────────
-
-    def test_no_session_returns_303_for_summarize(self, client, monkeypatch):
-        """无 session 返回 303。"""
-        monkeypatch.setattr(settings, "ADMIN_PASSWORD", "some-password")
-        resp = client.post("/admin/summarize", follow_redirects=False)
-        assert resp.status_code == 303
-
     def test_correct_session_batch_summarize(self, auth_client):
-        """已登录调用 batch summarize，mock 掉服务层。"""
-        with patch("app.routes.admin.summarize_batch", new_callable=AsyncMock) as mock:
-            mock.return_value = {
-                "status": "success",
-                "done": 0,
-                "failed": 0,
-                "total": 0,
-            }
+        """已登录调用 batch summarize，应创建后台任务。"""
+        with patch("app.routes.admin.enqueue_job"):
             resp = auth_client.post("/admin/summarize")
             assert resp.status_code == 200
-            assert resp.json()["status"] == "success"
-
-    def test_single_paper_not_found(self, auth_client):
-        """单篇总结不存在的论文返回 404。"""
-        from app.exceptions import NotFoundError
-
-        with patch(
-            "app.routes.admin.summarize_single",
-            new_callable=AsyncMock,
-            side_effect=NotFoundError("Paper not found: nonexistent.99999"),
-        ):
-            resp = auth_client.post("/admin/summarize/nonexistent.99999")
-            assert resp.status_code == 404
+            assert resp.json()["status"] == "queued"
+            assert "job_id" in resp.json()
 
 
 # ═══════════════════════════════════════════════════════════════════════
@@ -115,29 +84,12 @@ class TestAdminAuth:
 class TestAdminCrawl:
     """POST /admin/crawl 测试。"""
 
-    def test_crawl_default_today(self, auth_client):
-        """不指定日期时默认抓取今天。"""
-        with patch(
-            "app.routes.admin.run_crawl", new_callable=AsyncMock
-        ) as mock_crawl:
-            mock_crawl.return_value = {"found": 5, "new": 3, "status": "success"}
-            resp = auth_client.post("/admin/crawl")
-            assert resp.status_code == 200
-            data = resp.json()
-            assert data["status"] == "success"
-            mock_crawl.assert_called_once()
-
     def test_crawl_specific_date(self, auth_client):
         """指定日期抓取。"""
-        with patch(
-            "app.routes.admin.run_crawl", new_callable=AsyncMock
-        ) as mock_crawl:
-            mock_crawl.return_value = {"found": 2, "new": 1, "status": "success"}
+        with patch("app.routes.admin.enqueue_job"):
             resp = auth_client.post("/admin/crawl?date=2024-01-15")
             assert resp.status_code == 200
-            mock_crawl.assert_called_once()
-            call_args = mock_crawl.call_args
-            assert call_args[0][1] == "2024-01-15"
+            assert resp.json()["target_date"] == "2024-01-15"
 
 
 # ═══════════════════════════════════════════════════════════════════════
@@ -149,20 +101,20 @@ class TestAdminCleanup:
     """POST /admin/cleanup 测试。"""
 
     def test_cleanup_returns_stats(self, auth_client):
-        """清理应返回统计信息。"""
+        """同步清理排障接口应返回统计信息。"""
         with patch("app.routes.admin.cleanup_tmp") as mock_cleanup:
             mock_cleanup.return_value = {"scanned": 3, "removed": 1, "errors": []}
-            resp = auth_client.post("/admin/cleanup")
+            resp = auth_client.post("/admin/cleanup-now")
             assert resp.status_code == 200
             data = resp.json()
             assert data["scanned"] == 3
             assert data["removed"] == 1
 
     def test_cleanup_writes_log(self, auth_client, db_session):
-        """清理应写入 crawl_logs。"""
+        """同步清理排障接口应写入 crawl_logs。"""
         with patch("app.routes.admin.cleanup_tmp") as mock_cleanup:
             mock_cleanup.return_value = {"scanned": 0, "removed": 0, "errors": []}
-            auth_client.post("/admin/cleanup")
+            auth_client.post("/admin/cleanup-now")
 
         logs = (
             db_session.execute(select(CrawlLog).where(CrawlLog.task == "cleanup"))
@@ -195,19 +147,21 @@ class TestAdminDelete:
         assert resp.status_code == 422
 
     def test_delete_with_confirm(self, auth_client, db_session, sample_papers_range):
-        """confirm='DELETE' 时应执行删除。"""
-        resp = auth_client.post(
-            "/admin/delete",
-            json={
-                "date_start": "2024-01-10",
-                "date_end": "2024-01-12",
-                "include_notes": True,
-                "confirm": "DELETE",
-            },
-        )
+        """confirm='DELETE' 时应创建后台删除 job。"""
+        with patch("app.routes.admin.enqueue_job"):
+            resp = auth_client.post(
+                "/admin/delete",
+                json={
+                    "date_start": "2024-01-10",
+                    "date_end": "2024-01-12",
+                    "include_notes": True,
+                    "confirm": "DELETE",
+                },
+            )
         assert resp.status_code == 200
         data = resp.json()
-        assert data["deleted"] == 3
+        assert data["status"] == "queued"
+        assert db_session.get(Job, data["job_id"]) is not None
 
     def test_delete_invalid_date_range(self, auth_client):
         """date_start > date_end 应返回 400。"""
@@ -221,17 +175,6 @@ class TestAdminDelete:
         )
         assert resp.status_code == 400
 
-    def test_delete_without_confirm_field(self, auth_client):
-        """缺少 confirm 字段应返回 422。"""
-        resp = auth_client.post(
-            "/admin/delete",
-            json={
-                "date_start": "2024-01-10",
-                "date_end": "2024-01-12",
-            },
-        )
-        assert resp.status_code == 422
-
 
 # ═══════════════════════════════════════════════════════════════════════
 # Admin Routes — Logs
@@ -241,12 +184,6 @@ class TestAdminDelete:
 class TestAdminLogs:
     """GET /admin/logs 测试。"""
 
-    def test_logs_returns_page(self, auth_client):
-        """应返回管理日志页面。"""
-        resp = auth_client.get("/admin/logs")
-        assert resp.status_code == 200
-        assert "text/html" in resp.headers.get("content-type", "")
-
     def test_logs_requires_auth(self, client, monkeypatch):
         """日志页面需要鉴权。"""
         monkeypatch.setattr(settings, "ADMIN_PASSWORD", "some-password")
@@ -272,6 +209,126 @@ class TestAdminLogs:
         assert "crawl" in resp.text.lower() or "日志" in resp.text
 
 
+class TestAdminJobs:
+    """后台 job 查询接口测试。"""
+
+    def test_job_detail_returns_payload_and_events(self, auth_client, db_session):
+        """GET /admin/jobs/{id} 返回 job 主记录和事件。"""
+        with patch("app.routes.admin.enqueue_job"):
+            resp = auth_client.post("/admin/crawl?date=2024-01-15")
+        job_id = resp.json()["job_id"]
+
+        resp = auth_client.get(f"/admin/jobs/{job_id}")
+
+        assert resp.status_code == 200
+        data = resp.json()
+        assert data["id"] == job_id
+        assert data["type"] == "crawl_daily"
+        assert data["payload"] == {"target_date": "2024-01-15"}
+        assert data["events"][0]["stage"] == "created"
+
+    def test_job_detail_not_found(self, auth_client):
+        resp = auth_client.get("/admin/jobs/999999")
+        assert resp.status_code == 404
+
+
+class TestAdminSummaryStatus:
+    """总结状态管理接口测试。"""
+
+    def test_summary_status_json_filters_failed(
+        self, auth_client, db_session, sample_paper
+    ):
+        sample_paper.summary_status.status = SummaryState.FAILED
+        sample_paper.summary_status.retry_count = 2
+        sample_paper.summary_status.error_type = "timeout"
+        db_session.commit()
+
+        resp = auth_client.get("/admin/summary-status?status=failed")
+
+        assert resp.status_code == 200
+        data = resp.json()
+        assert data["total"] == 1
+        assert data["items"][0]["arxiv_id"] == sample_paper.arxiv_id
+        assert data["items"][0]["retry_count"] == 2
+
+    def test_retry_failed_resets_failed_statuses(
+        self, auth_client, db_session, sample_paper
+    ):
+        sample_paper.summary_status.status = SummaryState.PERMANENT_FAILURE
+        sample_paper.summary_status.error = "bad json"
+        sample_paper.summary_status.error_type = "json_invalid"
+        db_session.commit()
+
+        resp = auth_client.post("/admin/summary-retry-failed")
+
+        assert resp.status_code == 200
+        assert resp.json()["count"] == 1
+        db_session.refresh(sample_paper.summary_status)
+        assert sample_paper.summary_status.status == SummaryState.PENDING
+        assert sample_paper.summary_status.error is None
+        assert sample_paper.summary_status.error_type is None
+
+
+class TestAdminPapers:
+    """论文管理批量操作测试。"""
+
+    def test_single_delete_removes_paper_and_fts(
+        self, auth_client, db_session, sample_paper
+    ):
+        paper_id = sample_paper.id
+
+        resp = auth_client.post(f"/admin/paper-delete/{sample_paper.arxiv_id}")
+
+        assert resp.status_code == 200
+        assert db_session.get(type(sample_paper), paper_id) is None
+        fts_row = db_session.execute(
+            text("SELECT rowid FROM papers_fts WHERE rowid = :id"),
+            {"id": paper_id},
+        ).fetchone()
+        assert fts_row is None
+
+    def test_batch_delete_removes_papers_and_fts(
+        self, auth_client, db_session, sample_papers_range
+    ):
+        target_ids = [p.id for p in sample_papers_range[:2]]
+        target_arxiv_ids = [p.arxiv_id for p in sample_papers_range[:2]]
+
+        resp = auth_client.post(
+            "/admin/papers-batch-action",
+            json={"action": "delete", "arxiv_ids": target_arxiv_ids},
+        )
+
+        assert resp.status_code == 200
+        assert resp.json()["count"] == 2
+        remaining = db_session.execute(
+            text(
+                "SELECT rowid FROM papers_fts "
+                "WHERE rowid IN (:id1, :id2)"
+            ),
+            {"id1": target_ids[0], "id2": target_ids[1]},
+        ).fetchall()
+        assert remaining == []
+
+    def test_batch_summarize_sets_pending_status(
+        self, auth_client, db_session, sample_papers_range
+    ):
+        paper = sample_papers_range[0]
+        paper.summary_status.status = SummaryState.DONE
+        db_session.commit()
+
+        resp = auth_client.post(
+            "/admin/papers-batch-action",
+            json={"action": "summarize", "arxiv_ids": [paper.arxiv_id]},
+        )
+
+        assert resp.status_code == 200
+        status = db_session.scalar(
+            select(SummaryStatus).where(SummaryStatus.paper_id == paper.id)
+        )
+        assert status is not None
+        assert status.status == SummaryState.PENDING
+
+
 # ═══════════════════════════════════════════════════════════════════════
 # Scheduler 测试
 # ═══════════════════════════════════════════════════════════════════════
diff --git a/tests/test_crawler.py b/tests/test_crawler.py
index d663fb8..941b213 100644
--- a/tests/test_crawler.py
+++ b/tests/test_crawler.py
@@ -10,6 +10,7 @@ from app.services.crawler import (
     _parse_paper,
     crawl_daily,
     fetch_daily,
+    refresh_upvotes,
     upsert_papers,
 )
 
@@ -187,3 +188,62 @@ class TestCrawlDaily:
 
         assert result["status"] == "failed"
         assert "network error" in result["error"]
+
+
+class TestRefreshUpvotes:
+    @pytest.mark.asyncio
+    async def test_refresh_updates_existing_without_inserting_new(
+        self, db_session, sample_paper
+    ):
+        sample_paper.arxiv_id = "1706.03762"
+        sample_paper.upvotes = 10
+        db_session.commit()
+
+        with patch(
+            "app.services.crawler.fetch_daily",
+            new_callable=AsyncMock,
+            return_value=[
+                {
+                    "paper": {
+                        "id": "1706.03762",
+                        "upvotes": 999,
+                        "authors": [],
+                        "tags": [],
+                    }
+                },
+                {
+                    "paper": {
+                        "id": "2010.11929",
+                        "upvotes": 123,
+                        "authors": [],
+                        "tags": [],
+                    }
+                },
+            ],
+        ):
+            result = await refresh_upvotes(db_session, days=1)
+
+        db_session.refresh(sample_paper)
+        assert result["status"] == "success"
+        assert result["updated"] == 1
+        assert sample_paper.upvotes == 999
+        assert db_session.query(type(sample_paper)).count() == 1
+
+    @pytest.mark.asyncio
+    async def test_refresh_returns_partial_when_one_day_fails(self, db_session):
+        async def _fetch_daily(target_date):
+            if target_date.endswith("01"):
+                raise ConnectionError("hf down")
+            return []
+
+        with (
+            patch(
+                "app.services.crawler.recent_date_strs",
+                return_value=["2024-01-01", "2024-01-02"],
+            ),
+            patch("app.services.crawler.fetch_daily", side_effect=_fetch_daily),
+        ):
+            result = await refresh_upvotes(db_session, days=2)
+
+        assert result["status"] == "partial"
+        assert result["errors"] == ["2024-01-01: hf down"]
diff --git a/tests/test_derived.py b/tests/test_derived.py
new file mode 100644
index 0000000..769124b
--- /dev/null
+++ b/tests/test_derived.py
@@ -0,0 +1,80 @@
+"""派生索引维护测试。"""
+
+from __future__ import annotations
+
+from unittest.mock import patch
+
+from sqlalchemy import text
+
+from app.services.derived import reindex_chroma, reindex_fts
+
+
+class TestReindexFts:
+    def test_reindex_fts_rebuilds_missing_rows(self, db_session, sample_paper):
+        db_session.execute(
+            text("DELETE FROM papers_fts WHERE rowid = :id"),
+            {"id": sample_paper.id},
+        )
+        db_session.commit()
+
+        result = reindex_fts(db_session)
+
+        row = db_session.execute(
+            text("SELECT title_en, authors, tags FROM papers_fts WHERE rowid = :id"),
+            {"id": sample_paper.id},
+        ).fetchone()
+        assert result == {"status": "success", "indexed": 1}
+        assert row is not None
+        assert row[0] == sample_paper.title_en
+        assert "Alice Smith" in row[1]
+        assert "NLP" in row[2]
+
+    def test_reindex_fts_accepts_subset(self, db_session, sample_papers_range):
+        keep_id = sample_papers_range[0].id
+        skip_id = sample_papers_range[1].id
+        db_session.execute(text("DELETE FROM papers_fts"))
+        db_session.commit()
+
+        result = reindex_fts(db_session, paper_ids=[keep_id])
+
+        keep_row = db_session.execute(
+            text("SELECT rowid FROM papers_fts WHERE rowid = :id"),
+            {"id": keep_id},
+        ).fetchone()
+        skip_row = db_session.execute(
+            text("SELECT rowid FROM papers_fts WHERE rowid = :id"),
+            {"id": skip_id},
+        ).fetchone()
+        assert result["indexed"] == 1
+        assert keep_row is not None
+        assert skip_row is None
+
+
+class TestReindexChroma:
+    def test_reindex_chroma_indexes_only_summarized_papers(
+        self, db_session, sample_papers_with_summary
+    ):
+        with patch("app.services.embedder.index_paper", return_value=True) as mock_index:
+            result = reindex_chroma(db_session)
+
+        assert result["status"] == "success"
+        assert result["indexed"] == 4
+        assert mock_index.call_count == 4
+        indexed_ids = {call.args[0] for call in mock_index.call_args_list}
+        assert "2401.20001" in indexed_ids
+        assert "2401.20005" not in indexed_ids
+
+    def test_reindex_chroma_reports_partial_failures(
+        self, db_session, sample_papers_with_summary
+    ):
+        def _index_paper(arxiv_id, _texts):
+            if arxiv_id == "2401.20001":
+                raise RuntimeError("embedding failed")
+            return True
+
+        with patch("app.services.embedder.index_paper", side_effect=_index_paper):
+            result = reindex_chroma(db_session)
+
+        assert result["status"] == "partial"
+        assert result["indexed"] == 3
+        assert result["errors"] == ["2401.20001: embedding failed"]
diff --git a/tests/test_jobs.py b/tests/test_jobs.py
new file mode 100644
index 0000000..5b3be65
--- /dev/null
+++ b/tests/test_jobs.py
@@ -0,0 +1,111 @@
+"""后台 Job 服务测试。"""
+
+from __future__ import annotations
+
+from datetime import timedelta
+from unittest.mock import patch
+
+import pytest
+from sqlalchemy import select
+
+from app.models import Job, JobEvent, JobStatus, TaskLock
+from app.services.jobs import create_job, recover_stale_jobs, run_job
+from app.utils import utc_now
+
+
+class TestJobs:
+    def test_create_job_writes_event(self, db_session):
+        job = create_job(
+            db_session,
+            "cleanup_tmp",
+            owner="test",
+            payload={"reason": "unit-test"},
+        )
+
+        assert job.id is not None
+        assert job.status == JobStatus.QUEUED
+
+        events = (
+            db_session.execute(select(JobEvent).where(JobEvent.job_id == job.id))
+            .scalars()
+            .all()
+        )
+        assert len(events) == 1
+        assert events[0].stage == "created"
+
+    @pytest.mark.asyncio
+    async def test_run_job_success(self, db_session):
+        job = create_job(db_session, "cleanup_tmp", owner="test", payload={})
+
+        with patch("app.services.cleaner.cleanup_tmp") as mock_cleanup:
+            mock_cleanup.return_value = {"scanned": 1, "removed": 1, "errors": []}
+            result = await run_job(db_session, job.id)
+
+        refreshed = db_session.get(Job, job.id)
+        assert result["removed"] == 1
+        assert refreshed.status == JobStatus.SUCCESS
+        assert refreshed.result_json is not None
+
+    @pytest.mark.asyncio
+    async def test_run_job_failure_records_error(self, db_session):
+        job = create_job(db_session, "missing_job_type", owner="test", payload={})
+
+        result = await run_job(db_session, job.id)
+
+        refreshed = db_session.get(Job, job.id)
+        assert result["status"] == "failed"
+        assert refreshed.status == JobStatus.FAILED
+        assert "Unsupported job type" in refreshed.error
+
+    @pytest.mark.asyncio
+    async def test_run_job_dispatches_refresh_upvotes(self, db_session):
+        job = create_job(
+            db_session,
+            "refresh_upvotes",
+            owner="test",
+            payload={"days": 3},
+        )
+
+        with patch("app.services.crawler.refresh_upvotes") as mock_refresh:
+            mock_refresh.return_value = {"status": "success", "updated": 2}
+            result = await run_job(db_session, job.id)
+
+        mock_refresh.assert_awaited_once_with(db_session, days=3)
+        assert result["updated"] == 2
+
+    @pytest.mark.asyncio
+    async def test_run_job_dispatches_reindex_fts(self, db_session):
+        job = create_job(db_session, "reindex_fts", owner="test", payload={})
+
+        with patch("app.services.derived.reindex_fts") as mock_reindex:
+            mock_reindex.return_value = {"status": "success", "indexed": 5}
+            result = await run_job(db_session, job.id)
+
+        mock_reindex.assert_called_once_with(db_session)
+        assert result["indexed"] == 5
+
+    def test_recover_stale_jobs_and_locks(self, db_session):
+        old = utc_now() - timedelta(hours=7)
+        job = Job(
+            type="cleanup_tmp",
+            status=JobStatus.RUNNING,
+            owner="test",
+            created_at=old,
+            started_at=old,
+            heartbeat_at=old,
+        )
+        lock = TaskLock(
+            task="cleanup",
+            lock_key="tmp",
+            status="running",
+            owner="test",
+            acquired_at=old,
+        )
+        db_session.add_all([job, lock])
+        db_session.commit()
+
+        recovered = recover_stale_jobs(db_session)
+
+        assert recovered == 2
+        assert db_session.get(Job, job.id).status == JobStatus.STALE
+        assert db_session.get(TaskLock, lock.id).status == "stale"
diff --git a/tests/test_pages.py b/tests/test_pages.py
index 2516048..482cfcf 100644
--- a/tests/test_pages.py
+++ b/tests/test_pages.py
@@ -6,9 +6,6 @@ from datetime import date
 from unittest.mock import patch as upatch
 
 
-from app.config import settings
-
-
 # ═══════════════════════════════════════════════════════════════════════
 # Detail 页 & 相似论文
 # ═══════════════════════════════════════════════════════════════════════
@@ -37,29 +34,6 @@ class TestDetailPage:
 class TestTrendsDashboard:
     """趋势看板测试。"""
 
-    def test_trends_page_renders(self, client, sample_papers_with_summary):
-        """趋势看板页面正常渲染。"""
-        resp = client.get("/trends")
-        assert resp.status_code == 200
-        assert "趋势看板" in resp.text
-        assert "chart" in resp.text.lower() or "Chart" in resp.text
-
-    def test_trends_api_returns_data(self, client, sample_papers_with_summary):
-        """趋势 API 返回正确数据结构。"""
-        resp = client.get("/api/stats/trends")
-        assert resp.status_code == 200
-        data = resp.json()
-
-        assert "daily_counts" in data
-        assert "top_tags" in data
-        assert "upvotes_dist" in data
-        assert "summary_completion" in data
-
-        assert isinstance(data["daily_counts"], list)
-        assert isinstance(data["top_tags"], list)
-        assert isinstance(data["upvotes_dist"], list)
-        assert isinstance(data["summary_completion"], list)
-
     def test_trends_api_daily_counts(self, client, sample_papers_with_summary):
         """每日论文数量数据正确。"""
         # 使用测试数据的日期范围
@@ -108,12 +82,6 @@ class TestTrendsDashboard:
 class TestComparePage:
     """论文对比页测试。"""
 
-    def test_compare_page_no_ids(self, client):
-        """无 ID 时显示输入表单。"""
-        resp = client.get("/compare")
-        assert resp.status_code == 200
-        assert "对比" in resp.text
-
     def test_compare_page_with_ids(self, client, sample_papers_with_summary):
         """对比多篇论文正常渲染。"""
         resp = client.get("/compare?ids=2401.20001,2401.20002")
@@ -124,23 +92,6 @@ class TestComparePage:
         assert "一句话摘要" in resp.text
         assert "研究问题" in resp.text
 
-    def test_compare_page_max_5(self, client, sample_papers_with_summary):
-        """最多 5 篇。"""
-        ids = "2401.20001,2401.20002,2401.20003,2401.20004,2401.20005"
-        resp = client.get(f"/compare?ids={ids}")
-        assert resp.status_code == 200
-
-    def test_compare_page_over_5_truncates(self, client, sample_papers_with_summary):
-        """超过 5 篇截断。"""
-        ids = "2401.20001,2401.20002,2401.20003,2401.20004,2401.20005,2401.20006"
-        resp = client.get(f"/compare?ids={ids}")
-        assert resp.status_code == 200
-
-    def test_compare_page_invalid_ids(self, client):
-        """无效 ID 时显示空结果。"""
-        resp = client.get("/compare?ids=nonexistent.99999")
-        assert resp.status_code == 200
-
     def test_compare_page_shows_no_summary_placeholder(
         self, client, sample_papers_with_summary
     ):
@@ -149,65 +100,3 @@ class TestComparePage:
         resp = client.get("/compare?ids=2401.20005")
         assert resp.status_code == 200
         assert "暂无总结" in resp.text
-
-
-# ═══════════════════════════════════════════════════════════════════════
-# Nav Bar
-# ═══════════════════════════════════════════════════════════════════════
-
-
-class TestNavBar:
-    """导航栏测试。"""
-
-    def test_nav_includes_trends_link(self, client):
-        """导航栏应包含趋势链接。"""
-        resp = client.get("/search")
-        assert resp.status_code == 200
-        assert "/trends" in resp.text
-
-    def test_nav_includes_compare_implicitly(self, client):
-        """compare 页面可访问。"""
-        resp = client.get("/compare")
-        assert resp.status_code == 200
-
-
-# ═══════════════════════════════════════════════════════════════════════
-# Graceful Degradation（CHROMA_ENABLED=false）
-# ═══════════════════════════════════════════════════════════════════════
-
-
-class TestGracefulDegradation:
-    """CHROMA_ENABLED=false 时优雅降级测试。"""
-
-    def test_search_works_without_chroma(
-        self, client, monkeypatch, sample_papers_with_summary
-    ):
-        """CHROMA 关闭时 FTS5 搜索正常工作。"""
-        monkeypatch.setattr(settings, "CHROMA_ENABLED", False)
-        resp = client.get("/search?q=Test")
-        assert resp.status_code == 200
-        assert "Test Paper" in resp.text or "测试论文" in resp.text
-
-    def test_detail_works_without_chroma(
-        self, client, monkeypatch, sample_papers_with_summary
-    ):
-        """CHROMA 关闭时详情页正常工作。"""
-        monkeypatch.setattr(settings, "CHROMA_ENABLED", False)
-        resp = client.get("/paper/2401.20001")
-        assert resp.status_code == 200
-
-    def test_trends_works_without_chroma(
-        self, client, monkeypatch, sample_papers_with_summary
-    ):
-        """CHROMA 关闭时趋势看板正常工作。"""
-        monkeypatch.setattr(settings, "CHROMA_ENABLED", False)
-        resp = client.get("/trends")
-        assert resp.status_code == 200
-
-    def test_compare_works_without_chroma(
-        self, client, monkeypatch, sample_papers_with_summary
-    ):
-        """CHROMA 关闭时对比页正常工作。"""
-        monkeypatch.setattr(settings, "CHROMA_ENABLED", False)
-        resp = client.get("/compare?ids=2401.20001,2401.20002")
-        assert resp.status_code == 200
diff --git a/tests/test_searcher.py b/tests/test_searcher.py
index 861adfa..aa2ff0a 100644
--- a/tests/test_searcher.py
+++ b/tests/test_searcher.py
@@ -123,38 +123,12 @@ class TestSearchSemanticMode:
 class TestSearchRoutes:
     """搜索页面和 JSON API 路由测试。"""
 
-    def test_search_page_renders(self, client):
-        """GET /search 返回 200。"""
-        resp = client.get("/search")
-        assert resp.status_code == 200
-        assert "搜索" in resp.text
-
     def test_search_page_with_query(self, client, sample_paper):
         """GET /search?q=Test 返回搜索结果。"""
         resp = client.get("/search?q=Test")
         assert resp.status_code == 200
         assert "2401.12345" in resp.text
 
-    def test_search_page_with_tag(self, client, sample_paper):
-        """GET /search?tag=NLP 返回标签筛选结果。"""
-        resp = client.get("/search?tag=NLP")
-        assert resp.status_code == 200
-        assert "2401.12345" in resp.text
-
-    def test_search_page_keyword_mode(self, client, sample_papers_with_summary):
-        """搜索页 keyword 模式。"""
-        resp = client.get("/search?q=Test&mode=keyword")
-        assert resp.status_code == 200
-        assert "Test" in resp.text or "测试" in resp.text
-
-    def test_search_page_semantic_disabled(
-        self, client, monkeypatch, sample_papers_with_summary
-    ):
-        """语义模式 CHROMA_ENABLED=false 时仍能工作。"""
-        monkeypatch.setattr(settings, "CHROMA_ENABLED", False)
-        resp = client.get("/search?q=Test&mode=semantic")
-        assert resp.status_code == 200
-
     def test_search_api_json(self, client, sample_paper):
         """GET /api/search?q=Test 返回 JSON。"""
         resp = client.get("/api/search?q=Test")
@@ -170,14 +144,6 @@ class TestSearchRoutes:
         data = resp.json()
         assert data["total"] == 1
 
-    def test_search_api_with_mode(self, client, sample_papers_with_summary):
-        """搜索 API 支持 mode 参数。"""
-        resp = client.get("/api/search?q=Test&mode=keyword")
-        assert resp.status_code == 200
-        data = resp.json()
-        assert "results" in data
-        assert "total" in data
-
     def test_search_api_empty(self, client, sample_paper):
         """GET /api/search?q=nonexistent 返回空结果。"""
         resp = client.get("/api/search?q=nonexistent")
@@ -185,13 +151,6 @@ class TestSearchRoutes:
         data = resp.json()
         assert data["total"] == 0
 
-    def test_search_api_sort_by_date(self, client, sample_paper):
-        """GET /api/search?q=Test&sort=date 按日期排序。"""
-        resp = client.get("/api/search?q=Test&sort=date")
-        assert resp.status_code == 200
-        data = resp.json()
-        assert data["total"] >= 1
-
 
 # ═══════════════════════════════════════════════════════════════════════
 # Similar Paper API 测试
@@ -211,21 +170,6 @@ class TestSimilarAPI:
         data = resp.json()
         assert data["results"] == []
 
-    def test_similar_api_paper_not_found(self, client, monkeypatch):
-        """不存在的论文返回空。"""
-        monkeypatch.setattr(settings, "CHROMA_ENABLED", False)
-        resp = client.get("/api/similar/nonexistent.99999")
-        assert resp.status_code == 200
-        assert resp.json()["results"] == []
-
-    def test_similar_api_with_top_k(
-        self, client, monkeypatch, sample_papers_with_summary
-    ):
-        """top_k 参数控制返回数量。"""
-        monkeypatch.setattr(settings, "CHROMA_ENABLED", False)
-        resp = client.get("/api/similar/2401.20001?top_k=3")
-        assert resp.status_code == 200
-
 
 # ═══════════════════════════════════════════════════════════════════════
 # 阅读列表路由测试
@@ -235,12 +179,6 @@ class TestSimilarAPI:
 class TestReadingListRoute:
     """阅读列表页面测试。"""
 
-    def test_reading_list_empty(self, client):
-        """无收藏时显示空状态。"""
-        resp = client.get("/reading-list")
-        assert resp.status_code == 200
-        assert "阅读列表" in resp.text
-
     def test_reading_list_with_bookmark(self, client, sample_paper):
         """有收藏时显示论文。"""
         # 先收藏
@@ -302,13 +240,6 @@ class TestRssFeed:
         assert "<channel>" in resp.text
         assert "2401.12345" in resp.text
 
-    def test_rss_has_paper_item(self, client, sample_paper):
-        """RSS 包含论文条目。"""
-        resp = client.get("/rss.xml")
-        assert "<item>" in resp.text
-        assert "<title>" in resp.text
-        assert "/paper/2401.12345" in resp.text
-
     def test_rss_with_tag_filter(self, client, sample_paper):
         """GET /rss.xml?tag=NLP 按标签筛选。"""
         resp = client.get("/rss.xml?tag=NLP")