2026-05-14 19:55:12 +08:00
|
|
|
|
# LLM Intelligence Hub
|
|
|
|
|
|
|
|
|
|
|
|
面向 LLM 模型、定价与日报产出的情报采集项目，当前仓库提供：
|
|
|
|
|
|
|
|
|
|
|
|
- Go 采集脚本：采集 OpenRouter、多源补充数据与官方补录数据
|
|
|
|
|
|
- PostgreSQL 数据层：保存模型、区域定价、订阅套餐、日报与运行审计
|
|
|
|
|
|
- Go HTTP API：提供模型列表、套餐列表、最新正式日报入口
|
|
|
|
|
|
- Vite + React 前端：提供 Dashboard / Explorer 两个只读页面
|
|
|
|
|
|
- Shell 运维脚本：迁移、调度、备份、恢复、验收与性能门禁
|
|
|
|
|
|
|
2026-05-15 22:43:21 +08:00
|
|
|
|
## 两大模块
|
|
|
|
|
|
|
|
|
|
|
|
项目现在按运行职责拆成两大模块：
|
|
|
|
|
|
|
|
|
|
|
|
1. `情报采集与信号沉淀模块`
|
|
|
|
|
|
负责真实采集、官方补录、套餐导入、目录核验，以及把“新模型 / 价格变化 / 官方发布 / 活动窗口”等关键信号物化到 `daily_signal_snapshot`。
|
|
|
|
|
|
2. `日报与下游表达模块`
|
|
|
|
|
|
负责消费 `models`、`region_pricing`、`subscription_plan`、`daily_signal_snapshot` 等结构化事实，生成 HTML / Markdown 日报；后续视频、卡片流、推送等形态也应挂在这一层。
|
|
|
|
|
|
|
2026-05-14 19:55:12 +08:00
|
|
|
|
## 当前能力边界
|
|
|
|
|
|
|
|
|
|
|
|
- 真实生产主链路是“采集/导入脚本 + PostgreSQL + 日报生成器 + API Server + Nginx”
|
|
|
|
|
|
- 最新正式日报由 `scripts/run_daily.sh` 生成，并写入 `daily_report` / `report_runs`
|
|
|
|
|
|
- 手工复跑使用 `scripts/run_real_pipeline.sh`，不会把产物标记成正式日报
|
|
|
|
|
|
- 历史补跑使用 `scripts/rebuild_historical_report.sh YYYY-MM-DD`
|
|
|
|
|
|
- HTTP API 当前未内建认证、授权和限流；公网暴露前必须在网关层补齐
|
|
|
|
|
|
|
|
|
|
|
|
## 目录概览
|
|
|
|
|
|
|
|
|
|
|
|
```text
|
|
|
|
|
|
cmd/server/ Go API Server
|
|
|
|
|
|
internal/ 通用内部库（collector、retry）
|
|
|
|
|
|
scripts/ 采集、导入、日报、验收、运维脚本
|
|
|
|
|
|
db/migrations/ PostgreSQL 迁移
|
|
|
|
|
|
frontend/ Vite + React 前端
|
|
|
|
|
|
reports/daily/ 日报产物与归档
|
|
|
|
|
|
ops/ 运维配置（如 logrotate）
|
|
|
|
|
|
docs/ 补充说明与上线文档
|
|
|
|
|
|
```
|
|
|
|
|
|
|
|
|
|
|
|
## 本地启动
|
|
|
|
|
|
|
|
|
|
|
|
### 1. 准备环境
|
|
|
|
|
|
|
|
|
|
|
|
```bash
|
|
|
|
|
|
cp .env.example .env
|
|
|
|
|
|
```
|
|
|
|
|
|
|
|
|
|
|
|
至少需要配置：
|
|
|
|
|
|
|
|
|
|
|
|
- `DATABASE_URL`
|
|
|
|
|
|
- `OPENROUTER_API_KEY`（仅真实采集需要）
|
|
|
|
|
|
|
|
|
|
|
|
详细变量说明见 [docs/CONFIGURATION.md](docs/CONFIGURATION.md)。
|
|
|
|
|
|
|
|
|
|
|
|
### 2. 应用数据库迁移
|
|
|
|
|
|
|
|
|
|
|
|
```bash
|
|
|
|
|
|
bash scripts/apply_migration.sh
|
|
|
|
|
|
```
|
|
|
|
|
|
|
|
|
|
|
|
### 3. 启动 API Server
|
|
|
|
|
|
|
|
|
|
|
|
```bash
|
|
|
|
|
|
go run ./cmd/server
|
|
|
|
|
|
```
|
|
|
|
|
|
|
|
|
|
|
|
默认端口为 `8080`，可通过 `PORT` 覆盖。
|
|
|
|
|
|
|
|
|
|
|
|
### 4. 启动前端开发环境
|
|
|
|
|
|
|
|
|
|
|
|
```bash
|
|
|
|
|
|
cd frontend
|
|
|
|
|
|
npm install
|
|
|
|
|
|
npm run dev
|
|
|
|
|
|
```
|
|
|
|
|
|
|
|
|
|
|
|
## 生产运行主链路
|
|
|
|
|
|
|
2026-05-15 22:43:21 +08:00
|
|
|
|
### 第一模块独立运行
|
|
|
|
|
|
|
|
|
|
|
|
```bash
|
|
|
|
|
|
bash scripts/run_intel_pipeline.sh
|
|
|
|
|
|
```
|
|
|
|
|
|
|
|
|
|
|
|
该入口只执行第一模块：
|
|
|
|
|
|
|
|
|
|
|
|
1. 真实采集与多源补充
|
|
|
|
|
|
2. 官方模型价格与套餐导入
|
|
|
|
|
|
3. 平台目录核验
|
|
|
|
|
|
4. 每日关键信号物化到 `daily_signal_snapshot`
|
|
|
|
|
|
|
|
|
|
|
|
它不会生成日报，适合先把“数据与信号层”单独跑通。
|
|
|
|
|
|
|
2026-05-14 19:55:12 +08:00
|
|
|
|
### 正式日报调度
|
|
|
|
|
|
|
|
|
|
|
|
```bash
|
|
|
|
|
|
bash scripts/run_daily.sh
|
|
|
|
|
|
```
|
|
|
|
|
|
|
|
|
|
|
|
该脚本负责：
|
|
|
|
|
|
|
|
|
|
|
|
1. OpenRouter 真实采集
|
|
|
|
|
|
2. 多源补充同步
|
|
|
|
|
|
3. 官方导入脚本执行
|
2026-05-15 22:43:21 +08:00
|
|
|
|
4. 每日关键信号物化
|
|
|
|
|
|
5. 数据质量检查
|
|
|
|
|
|
6. Markdown / HTML 日报生成
|
|
|
|
|
|
7. 日报归档
|
|
|
|
|
|
8. `daily_report` / `report_runs` 审计写入
|
|
|
|
|
|
9. 失败时降级复制昨日报告并可选飞书告警
|
2026-05-14 19:55:12 +08:00
|
|
|
|
|
|
|
|
|
|
### 手工真实复跑
|
|
|
|
|
|
|
|
|
|
|
|
```bash
|
|
|
|
|
|
bash scripts/run_real_pipeline.sh
|
|
|
|
|
|
```
|
|
|
|
|
|
|
|
|
|
|
|
适用于联调、排障、上线后人工验证。该入口写入：
|
|
|
|
|
|
|
|
|
|
|
|
- `run_kind=manual`
|
|
|
|
|
|
- `trigger_source=pipeline`
|
|
|
|
|
|
- `is_official_daily=false`
|
|
|
|
|
|
|
|
|
|
|
|
### 历史补跑
|
|
|
|
|
|
|
|
|
|
|
|
```bash
|
|
|
|
|
|
bash scripts/rebuild_historical_report.sh 2026-05-13
|
|
|
|
|
|
```
|
|
|
|
|
|
|
|
|
|
|
|
该入口写入：
|
|
|
|
|
|
|
|
|
|
|
|
- `run_kind=historical_rebuild`
|
|
|
|
|
|
- `trigger_source=rebuild_script`
|
|
|
|
|
|
- `is_official_daily=false`
|
|
|
|
|
|
|
|
|
|
|
|
## 常用命令
|
|
|
|
|
|
|
|
|
|
|
|
```bash
|
|
|
|
|
|
go test ./...
|
|
|
|
|
|
bash scripts/test.sh
|
|
|
|
|
|
bash scripts/verify_pre_phase6.sh
|
|
|
|
|
|
bash scripts/verify_phase6.sh
|
|
|
|
|
|
bash healthcheck.sh
|
|
|
|
|
|
cd frontend && npm run test -- --run
|
|
|
|
|
|
cd frontend && npm run build
|
|
|
|
|
|
```
|
|
|
|
|
|
|
|
|
|
|
|
## API 概览
|
|
|
|
|
|
|
|
|
|
|
|
- `GET /health`
|
|
|
|
|
|
- `GET /api/v1/models`
|
|
|
|
|
|
- `GET /api/v1/subscription-plans`
|
|
|
|
|
|
- `GET /api/v1/reports/latest`
|
|
|
|
|
|
- `GET /api/v1/reports/latest/markdown`
|
|
|
|
|
|
- `GET /api/v1/reports/latest/html`
|
|
|
|
|
|
|
|
|
|
|
|
完整字段与示例见 [docs/API_REFERENCE.md](docs/API_REFERENCE.md)。
|
|
|
|
|
|
|
|
|
|
|
|
## 文档索引
|
|
|
|
|
|
|
|
|
|
|
|
- [docs/CONFIGURATION.md](docs/CONFIGURATION.md)：环境变量、运行语义、配置约束
|
|
|
|
|
|
- [docs/API_REFERENCE.md](docs/API_REFERENCE.md)：API 入口、返回体与排障说明
|
|
|
|
|
|
- [docs/PRODUCTION_CHECKLIST.md](docs/PRODUCTION_CHECKLIST.md)：生产上线前检查、发布与回滚流程
|
|
|
|
|
|
- [DEPLOYMENT.md](DEPLOYMENT.md)：部署步骤与快速启动
|
|
|
|
|
|
- [RUNBOOK.md](RUNBOOK.md)：运维巡检、故障排查、备份恢复
|
|
|
|
|
|
- [TECHNICAL_DESIGN.md](TECHNICAL_DESIGN.md)：详细技术设计与数据模型演进背景
|
|
|
|
|
|
- [docs/PERFORMANCE_TEST.md](docs/PERFORMANCE_TEST.md)：性能基线
|
|
|
|
|
|
|
|
|
|
|
|
## 生产上线最低门禁
|
|
|
|
|
|
|
|
|
|
|
|
建议把以下检查作为发布前硬门禁：
|
|
|
|
|
|
|
|
|
|
|
|
```bash
|
|
|
|
|
|
bash scripts/verify_pre_phase6.sh
|
|
|
|
|
|
bash scripts/verify_phase6.sh
|
|
|
|
|
|
```
|
|
|
|
|
|
|
|
|
|
|
|
上线后首轮冒烟建议至少覆盖：
|
|
|
|
|
|
|
|
|
|
|
|
```bash
|
|
|
|
|
|
curl -fsS http://127.0.0.1:8080/health
|
|
|
|
|
|
curl -fsS http://127.0.0.1:8080/api/v1/models
|
|
|
|
|
|
curl -fsS http://127.0.0.1:8080/api/v1/reports/latest
|
|
|
|
|
|
```
|
