60b062daf2
- backend: FastAPI + SQLAlchemy 2.0(async) + asyncpg + Alembic - 7 API routes: auth/me/articles/sources/bookmarks/subscriptions/admin - models: User/Source/Article/Bookmark/Subscription/ApiToken - services: RSS fetcher (feedparser) + Tencent TMT translator with quota + cache + local NLLB fallback - workers: APScheduler + asyncio pipeline (fetch -> dedupe -> insert -> translate) - seed scripts: create_user, seed_sources (5 RSS: Reuters/BBC/Al Jazeera/NHK/DW) - frontend: Vue 3 + Vite + Naive UI + Pinia + vue-router - pages: Login, Feed (24h), ArticleDetail, Sources, Bookmarks, AdminSources - deploy: docker-compose (postgres/redis/api/worker/frontend/caddy) - docs: README, DEPLOY, architecture, acceptance
78 lines
2.3 KiB
Markdown
78 lines
2.3 KiB
Markdown
# 架构设计
|
||
|
||
> 对应方案 v0.1 的实现版本。
|
||
|
||
## 模块边界
|
||
|
||
| 模块 | 路径 | 职责 |
|
||
| --- | --- | --- |
|
||
| API | `backend/app/api/` | HTTP 路由,处理鉴权 / 入参 / 出参 |
|
||
| 业务 | `backend/app/services/` | 抓取、翻译、领域逻辑 |
|
||
| Worker | `backend/app/workers/` | 后台调度、pipeline |
|
||
| 数据 | `backend/app/models/` | SQLAlchemy ORM |
|
||
| 迁移 | `backend/alembic/` | 数据库 schema 版本 |
|
||
| 前端 | `frontend/src/` | Vue 3 + Naive UI |
|
||
|
||
## 数据流
|
||
|
||
```
|
||
Source (DB)
|
||
│
|
||
▼
|
||
Scheduler (cron / interval)
|
||
│
|
||
▼
|
||
RSSFetcher.fetch() ── HTTP GET ──► upstream RSS
|
||
│
|
||
▼
|
||
FetchedItem list
|
||
│
|
||
▼ (url_hash UNIQUE 去重)
|
||
Article INSERT
|
||
│
|
||
▼ (translation_status='pending')
|
||
TranslationService.translate()
|
||
├─ Redis cache hit → return
|
||
├─ quota check
|
||
├─ Tencent TMT (主) ──► 30 天 Redis 月度计数
|
||
└─ Local NLLB (降级,需启用)
|
||
│
|
||
▼
|
||
Article UPDATE (title_zh / body_zh_* / status)
|
||
```
|
||
|
||
## 关键设计决策
|
||
|
||
- **PostgreSQL**:UNIQUE 约束 + `ON CONFLICT DO NOTHING` 做去重,O(1) 写
|
||
- **Redis 三用**:翻译缓存(30 天 TTL)+ 月度配额(INCRBY)+ 后续限流
|
||
- **AScheduler 重构 jobs**:每天 00:30 从 DB 重新读,运行时新增源自动生效
|
||
- **翻译分块**:按段落切,单段 > 1500 字符按句号再切,单请求 ≤ 4500 字符(腾讯 TMT 上限)
|
||
- **失败退避**:某源连续失败 3 次,fetch_interval × 2(封顶 720 分钟),成功一次恢复
|
||
- **API Token 双轨**:网页用短期 JWT(15min)+ refresh(14d);Android 用长期 API Token(可独立撤销)
|
||
|
||
## 字段保留
|
||
|
||
`articles` 表里这些字段已建,MVP 全部 null,后续 enrichment 阶段直接写值不动表:
|
||
- `category` / `commentary` / `entities` / `sentiment` / `topic_id` / `bias`
|
||
|
||
## 安全
|
||
|
||
- 密码 bcrypt(cost=12)
|
||
- JWT 走 HTTPS-only cookie(网页) / Bearer header(APP)
|
||
- 数据库/Redis 不暴露到宿主机
|
||
- Caddy 做 TLS 终止
|
||
- API 限流(MVP 暂未实现,后续加)
|
||
|
||
## 不在 MVP
|
||
|
||
- ❌ 全文搜索(可用 PG `to_tsvector`,MVP 先简单 ILIKE)
|
||
- ❌ PWA 离线缓存
|
||
- ❌ Android 客户端
|
||
- ❌ 自动分类/点评/实体识别
|
||
- ❌ 主题聚类
|
||
- ❌ 跨源立场
|
||
- ❌ Telegram 推送
|
||
- ❌ i18n(只 zh)
|
||
|
||
见 [`DEPLOY.md`](../DEPLOY.md) 跑起来,见 [`../README.md`](../README.md) 看全貌。
|