autobot-review

内容来源：README.md（说明文档） · 原始地址 · 查看安装指南

原始内容

# ReviewBot - AI-Powered GitLab Code Review System

[![NestJS](https://img.shields.io/badge/NestJS-E0234E?logo=nestjs&logoColor=white)](https://nestjs.com/)
[![React](https://img.shields.io/badge/React-61DAFB?logo=react&logoColor=black)](https://reactjs.org/)
[![TypeScript](https://img.shields.io/badge/TypeScript-3178C6?logo=typescript&logoColor=white)](https://www.typescriptlang.org/)
[![PostgreSQL](https://img.shields.io/badge/PostgreSQL-4169E1?logo=postgresql&logoColor=white)](https://www.postgresql.org/)
[![Azure OpenAI](https://img.shields.io/badge/Azure%20OpenAI-0078D4?logo=microsoft-azure&logoColor=white)](https://azure.microsoft.com/en-us/products/ai-services/openai-service)

Automated code review bot for GitLab using Azure OpenAI Haiku 4.5. Provides intelligent code analysis, security checks, performance suggestions, and inline comments directly on merge requests.

## 📸 Screenshots

### Dashboard
![Dashboard Overview](https://github.com/binhmuc/autobot-review/raw/HEAD/images/dashboard.png)
*Real-time metrics showing code quality scores, issue breakdown, and review history*

### GitLab Integration
![Inline Comments](https://github.com/binhmuc/autobot-review/raw/HEAD/images/inline-comments.jpeg)
*AI-powered inline comments directly on merge request code changes*

![MR Summary](https://github.com/binhmuc/autobot-review/raw/HEAD/images/mr-summary.jpeg)
*Comprehensive merge request summary with quality score and categorized issues*

### Developer Analytics
![Developer Metrics](https://github.com/binhmuc/autobot-review/raw/HEAD/images/developer-metrics.png)
*Track individual developer performance and improvement trends over time*

> **Note**: Add your screenshots to the `/images` folder to display them here.

## 🌟 Features

### Code Review Intelligence
- **AI-Powered Analysis**: Utilizes Azure OpenAI Haiku 4.5 for fast, accurate code reviews
- **Multi-Language Support**: Analyzes TypeScript, JavaScript, Python, Java, Go, Rust, PHP, and more
- **Smart Batching**: Combines multiple file changes into single LLM call for efficiency (≤500 lines)
- **False Positive Filtering**: Verifies LLM findings against actual file content to reduce noise

### Review Capabilities
- **Inline Comments**: Posts comments directly on specific code lines for critical/high/medium issues
- **Comprehensive Summary**: Detailed MR-level summary with quality scores and issue breakdown
- **Issue Categorization**:
  - **Severity**: Critical, High, Medium, Low
  - **Type**: Security, Performance, Logic, Style
- **Context-Aware**: Includes ±10 lines of context and imports for accurate analysis

### Developer Metrics
- **Quality Scoring**: 0-100 score based on issue severity
- **Historical Tracking**: Review history and trends per developer
- **Project Analytics**: Aggregate metrics across projects

### Integration
- **GitLab Webhooks**: Automatic review on MR open/update
- **Queue Processing**: Async job processing with Bull + Redis
- **API Access**: RESTful API for frontend consumption

## 🏗️ Architecture

```
reviewbot/
├── reviewbot-backend/     # NestJS API + Queue Processor
│   ├── src/
│   │   ├── webhook/       # GitLab webhook receiver
│   │   ├── llm/           # Azure OpenAI integration
│   │   ├── queue/         # Bull queue processing
│   │   ├── gitlab/        # GitLab API client
│   │   └── prisma/        # Database models
│   ├── prisma/            # Database schema
│   └── docker-compose.yml # Infrastructure stack
│
└── reviewbot-frontend/    # React + Vite Dashboard
    └── src/
        ├── pages/         # Dashboard, Projects, Developers, Reviews
        ├── components/    # Reusable UI components
        └── api/           # API client

```

### Tech Stack

**Backend**:
- NestJS (Node.js framework)
- Prisma ORM
- PostgreSQL (database)
- Redis + Bull (queue processing)
- Azure OpenAI API (Haiku 4.5)
- GitLab REST API (@gitbeaker/rest)

**Frontend**:
- React 19
- Vite (build tool)
- TanStack Query (data fetching)
- Tailwind CSS + shadcn/ui
- Recharts (visualization)

## 📋 Prerequisites

- **Node.js**: v18+ (backend and frontend)
- **Docker & Docker Compose**: For PostgreSQL and Redis
- **Azure OpenAI**: API key and Haiku 4.5 deployment
- **GitLab**: Personal access token with API scope
- **Git**: For version control

## 🚀 Quick Start

### 1. Clone Repository

```bash
git clone <your-repo-url>
cd reviewbot
```

### 2. Backend Setup

```bash
cd reviewbot-backend

# Install dependencies
npm install

# Copy environment template
cp .env.example .env

# Edit .env with your credentials
nano .env
```

**Required Environment Variables**:
```env
# Database
DATABASE_URL=postgresql://reviewbot:your_password@localhost:5432/reviewbot
DB_PASSWORD=your_password

# Redis
REDIS_HOST=localhost
REDIS_PORT=6379

# Azure OpenAI
AZURE_OPENAI_KEY=your_api_key
AZURE_OPENAI_ENDPOINT=https://your-resource.cognitiveservices.azure.com/
AZURE_OPENAI_DEPLOYMENT=haiku-4-5
AZURE_OPENAI_MODEL_NAME=haiku-4-5

# GitLab
GITLAB_HOST=https://gitlab.com
GITLAB_WEBHOOK_SECRET=your_webhook_secret
GITLAB_ACCESS_TOKEN=your_gitlab_token

# Auth
JWT_SECRET=your_jwt_secret
ADMIN_USERNAME=admin
ADMIN_PASSWORD=change_me
```

### 3. Start Infrastructure

```bash
# Start PostgreSQL and Redis
docker-compose up -d postgres redis

# Run database migrations
npm run prisma:generate
npm run prisma:migrate
```

### 4. Start Backend

```bash
# Development mode with hot-reload
npm run start:dev

# Production mode
npm run build
npm run start:prod
```

Backend runs on `http://localhost:3000`

### 5. Frontend Setup

```bash
cd ../reviewbot-frontend

# Install dependencies
npm install

# Create .env.local
echo "VITE_API_URL=http://localhost:3000" > .env.local

# Start development server
npm run dev
```

Frontend runs on `http://localhost:5173`

## ⚙️ Configuration

### GitLab Webhook Setup

1. Go to your GitLab project → **Settings** → **Webhooks**
2. Add webhook:
   - **URL**: `http://your-server:3000/webhook`
   - **Secret Token**: Use value from `GITLAB_WEBHOOK_SECRET`
   - **Trigger**: ✅ Merge request events
   - **SSL verification**: Enable if using HTTPS
3. Click **Add webhook**
4. Test by creating a test MR

### Azure OpenAI Setup

1. Create Azure OpenAI resource in Azure Portal
2. Deploy **Haiku 4.5** model (or your preferred model)
3. Get **API key** and **endpoint** from Keys and Endpoint section
4. Update `.env` with credentials

## 📊 Usage

### Dashboard

Access the frontend at `http://localhost:5173` to view:

- **Code Quality Metrics**: Issues found, review scores, trends
- **Project Management**: Configure webhooks, view project stats
- **Developer Performance**: Individual metrics, improvement trends
- **Review History**: Full review details with inline comments

### API Endpoints

**Webhook** (GitLab calls this):
- `POST /webhook` - Receive MR events

**Projects**:
- `GET /projects` - List all projects
- `POST /projects` - Add project
- `GET /projects/:id/reviews` - Project review history

**Reviews**:
- `GET /reviews` - List all reviews
- `GET /reviews/:id` - Review details

**Developers**:
- `GET /developers` - List all developers
- `GET /developers/:id/stats` - Developer metrics

### Review Workflow

1. **Developer creates/updates MR** → GitLab webhook fires
2. **Backend receives webhook** → Creates review job in queue
3. **Queue processor**:
   - Fetches MR diffs from GitLab
   - Collects file context (imports, surrounding lines)
   - Decides batching strategy (≤500 lines = batch)
   - Calls Azure OpenAI for review
   - Verifies issues to filter false positives
   - Posts inline comments (critical/high/medium)
   - Posts summary comment with all issues
4. **Frontend displays**:
   - Summary comment appears on MR
   - Inline comments appear on specific lines
   - Dashboard shows updated metrics

## 🔧 Development

### Project Structure

```
reviewbot-backend/src/
├── main.ts                # Application entry point
├── webhook/
│   ├── webhook.controller.ts    # Webhook endpoint
│   └── webhook.service.ts       # Webhook validation
├── llm/
│   ├── llm.service.ts           # Azure OpenAI integration
│   ├── diff-processor.ts        # Diff parsing
│   └── issue-verifier.service.ts # False positive filter
├── queue/
│   └── review-processor.ts      # Bull queue job handler
├── gitlab/
│   └── gitlab.service.ts        # GitLab API client
└── prisma/
    └── prisma.service.ts        # Database client
```

### Database Schema

```prisma
model Project {
  id              String   @id @default(cuid())
  gitlabProjectId Int      @unique
  name            String
  webhookUrl      String?
  reviews         Review[]
}

model Developer {
  id           String   @id @default(cuid())
  gitlabUserId Int      @unique
  username     String
  email        String?
  reviews      Review[]
}

model Review {
  id              String       @id @default(cuid())
  mergeRequestIid Int
  projectId       String
  developerId     String
  reviewContent   Json
  qualityScore    Int
  issuesFound     Int
  status          ReviewStatus
  createdAt       DateTime     @default(now())
  project         Project      @relation(...)
  developer       Developer    @relation(...)
}
```

### Running Tests

```bash
# Backend tests
cd reviewbot-backend
npm test

# Frontend tests
cd reviewbot-frontend
npm run test
```

### Building for Production

```bash
# Backend
cd reviewbot-backend
npm run build

# Frontend
cd reviewbot-frontend
npm run build
```

## 🐳 Docker Deployment

### Option 1: Docker Compose (Full Stack)

```bash
cd reviewbot-backend

# Build and start all services
docker-compose up -d

# View logs
docker-compose logs -f

# Stop services
docker-compose down
```

Services:
- **PostgreSQL**: `localhost:5432`
- **Redis**: `localhost:6379`
- **Backend**: `localhost:3000`
- **Frontend**: `localhost:5173`

### Option 2: Separate Services

```bash
# Start only infrastructure
docker-compose up -d postgres redis

# Run backend locally
npm run start:dev

# Run frontend locally
cd ../reviewbot-frontend
npm run dev
```

## 🔍 Monitoring & Debugging

### Logs

```bash
# Backend logs (development)
npm run start:dev

# Docker logs
docker-compose logs -f backend

# Redis queue monitoring
npm run prisma:studio
# Navigate to Review model to see job status
```

### Common Issues

**Inline comments not posting**:
- Verify `GITLAB_ACCESS_TOKEN` has `api` scope
- Check `baseSha`, `headSha`, `startSha` are valid
- Ensure line number is within changed lines in diff

**Queue jobs stuck**:
- Check Redis connection: `redis-cli ping`
- Verify `REDIS_HOST` and `REDIS_PORT` in `.env`
- Check Bull queue in Prisma Studio

**LLM timeout**:
- Verify Azure OpenAI credentials
- Check deployment name matches `AZURE_OPENAI_DEPLOYMENT`
- Increase timeout in `llm.service.ts` if needed

## 🛠️ Advanced Configuration

### Batching Threshold

Edit `reviewbot-backend/src/queue/review-processor.ts`:

```typescript
// Line 139: Change threshold for batching
const shouldBatch = totalChangedLines <= 500 && allChunksWithContext.length > 1;
//                                       ^^^
//                                       Adjust this value
```

### Context Lines

Edit `reviewbot-backend/src/queue/review-processor.ts`:

```typescript
// Line 95: Adjust context lines before/after changes
const chunks = this.diffProcessor.extractChangedLinesWithContext(
  diff.diff,
  10, // Change this value (default: 10)
);
```

### File Limit

Edit `reviewbot-backend/src/queue/review-processor.ts`:

```typescript
// Line 19: Maximum files per review
private readonly MAX_FILES = 50; // Adjust to prevent token overflow
```

## 📝 API Documentation

### Swagger/OpenAPI

Access interactive API documentation at:
```
http://localhost:3000/api/docs
```

## 🤝 Contributing

Contributions welcome! Please follow these steps:

1. Fork the repository
2. Create feature branch: `git checkout -b feature/amazing-feature`
3. Commit changes: `git commit -m 'Add amazing feature'`
4. Push to branch: `git push origin feature/amazing-feature`
5. Open Pull Request

## 📄 License

This project is licensed under the ISC License.

## 🙏 Acknowledgments

- **Azure OpenAI**: Haiku 4.5 model for fast code analysis
- **GitLab**: Webhook integration and API
- **NestJS**: Robust backend framework
- **Prisma**: Type-safe database ORM
- **shadcn/ui**: Beautiful UI components

## 📧 Support

For issues, questions, or contributions:
- Open an issue on GitHub
- Check existing documentation in `/docs`
- Review `/plans` for project roadmap

---

**Made with ❤️ for better code quality**

ReviewBot - AI-Powered GitLab Code Review System

Automated code review bot for GitLab using Azure OpenAI Haiku 4.5. Provides intelligent code analysis, security checks, performance suggestions, and inline comments directly on merge requests.

📸 Screenshots

Dashboard

Dashboard Overview Real-time metrics showing code quality scores, issue breakdown, and review history

GitLab Integration

AI-powered inline comments directly on merge request code changes

MR Summary Comprehensive merge request summary with quality score and categorized issues

Developer Analytics

Developer Metrics Track individual developer performance and improvement trends over time

Note: Add your screenshots to the /images folder to display them here.

🌟 Features

Code Review Intelligence

AI-Powered Analysis: Utilizes Azure OpenAI Haiku 4.5 for fast, accurate code reviews
Multi-Language Support: Analyzes TypeScript, JavaScript, Python, Java, Go, Rust, PHP, and more
Smart Batching: Combines multiple file changes into single LLM call for efficiency (≤500 lines)
False Positive Filtering: Verifies LLM findings against actual file content to reduce noise

Review Capabilities

Inline Comments: Posts comments directly on specific code lines for critical/high/medium issues
Comprehensive Summary: Detailed MR-level summary with quality scores and issue breakdown
Issue Categorization:
- Severity: Critical, High, Medium, Low
- Type: Security, Performance, Logic, Style
Context-Aware: Includes ±10 lines of context and imports for accurate analysis

Developer Metrics

Quality Scoring: 0-100 score based on issue severity
Historical Tracking: Review history and trends per developer
Project Analytics: Aggregate metrics across projects

Integration

GitLab Webhooks: Automatic review on MR open/update
Queue Processing: Async job processing with Bull + Redis
API Access: RESTful API for frontend consumption

🏗️ Architecture

reviewbot/
├── reviewbot-backend/     # NestJS API + Queue Processor
│   ├── src/
│   │   ├── webhook/       # GitLab webhook receiver
│   │   ├── llm/           # Azure OpenAI integration
│   │   ├── queue/         # Bull queue processing
│   │   ├── gitlab/        # GitLab API client
│   │   └── prisma/        # Database models
│   ├── prisma/            # Database schema
│   └── docker-compose.yml # Infrastructure stack
│
└── reviewbot-frontend/    # React + Vite Dashboard
    └── src/
        ├── pages/         # Dashboard, Projects, Developers, Reviews
        ├── components/    # Reusable UI components
        └── api/           # API client

Tech Stack

Backend:

NestJS (Node.js framework)
Prisma ORM
PostgreSQL (database)
Redis + Bull (queue processing)
Azure OpenAI API (Haiku 4.5)
GitLab REST API (@gitbeaker/rest)

Frontend:

React 19
Vite (build tool)
TanStack Query (data fetching)
Tailwind CSS + shadcn/ui
Recharts (visualization)

📋 Prerequisites

Node.js: v18+ (backend and frontend)
Docker & Docker Compose: For PostgreSQL and Redis
Azure OpenAI: API key and Haiku 4.5 deployment
GitLab: Personal access token with API scope
Git: For version control

🚀 Quick Start

1. Clone Repository

git clone <your-repo-url>
cd reviewbot

2. Backend Setup

cd reviewbot-backend

# Install dependencies
npm install

# Copy environment template
cp .env.example .env

# Edit .env with your credentials
nano .env

Required Environment Variables:

# Database
DATABASE_URL=postgresql://reviewbot:your_password@localhost:5432/reviewbot
DB_PASSWORD=your_password

# Redis
REDIS_HOST=localhost
REDIS_PORT=6379

# Azure OpenAI
AZURE_OPENAI_KEY=your_api_key
AZURE_OPENAI_ENDPOINT=https://your-resource.cognitiveservices.azure.com/
AZURE_OPENAI_DEPLOYMENT=haiku-4-5
AZURE_OPENAI_MODEL_NAME=haiku-4-5

# GitLab
GITLAB_HOST=https://gitlab.com
GITLAB_WEBHOOK_SECRET=your_webhook_secret
GITLAB_ACCESS_TOKEN=your_gitlab_token

# Auth
JWT_SECRET=your_jwt_secret
ADMIN_USERNAME=admin
ADMIN_PASSWORD=change_me

3. Start Infrastructure

# Start PostgreSQL and Redis
docker-compose up -d postgres redis

# Run database migrations
npm run prisma:generate
npm run prisma:migrate

4. Start Backend

# Development mode with hot-reload
npm run start:dev

# Production mode
npm run build
npm run start:prod

Backend runs on http://localhost:3000

5. Frontend Setup

cd ../reviewbot-frontend

# Install dependencies
npm install

# Create .env.local
echo "VITE_API_URL=http://localhost:3000" > .env.local

# Start development server
npm run dev

Frontend runs on http://localhost:5173

⚙️ Configuration

GitLab Webhook Setup

Go to your GitLab project → Settings → Webhooks
Add webhook:
- URL: http://your-server:3000/webhook
- Secret Token: Use value from GITLAB_WEBHOOK_SECRET
- Trigger: ✅ Merge request events
- SSL verification: Enable if using HTTPS
Click Add webhook
Test by creating a test MR

Azure OpenAI Setup

Create Azure OpenAI resource in Azure Portal
Deploy Haiku 4.5 model (or your preferred model)
Get API key and endpoint from Keys and Endpoint section
Update .env with credentials

📊 Usage

Dashboard

Access the frontend at http://localhost:5173 to view:

Code Quality Metrics: Issues found, review scores, trends
Project Management: Configure webhooks, view project stats
Developer Performance: Individual metrics, improvement trends
Review History: Full review details with inline comments

API Endpoints

Webhook (GitLab calls this):

POST /webhook - Receive MR events

Projects:

GET /projects - List all projects
POST /projects - Add project
GET /projects/:id/reviews - Project review history

Reviews:

GET /reviews - List all reviews
GET /reviews/:id - Review details

Developers:

GET /developers - List all developers
GET /developers/:id/stats - Developer metrics

Review Workflow

Developer creates/updates MR → GitLab webhook fires
Backend receives webhook → Creates review job in queue
Queue processor:
- Fetches MR diffs from GitLab
- Collects file context (imports, surrounding lines)
- Decides batching strategy (≤500 lines = batch)
- Calls Azure OpenAI for review
- Verifies issues to filter false positives
- Posts inline comments (critical/high/medium)
- Posts summary comment with all issues
Frontend displays:
- Summary comment appears on MR
- Inline comments appear on specific lines
- Dashboard shows updated metrics

🔧 Development

Project Structure

reviewbot-backend/src/
├── main.ts                # Application entry point
├── webhook/
│   ├── webhook.controller.ts    # Webhook endpoint
│   └── webhook.service.ts       # Webhook validation
├── llm/
│   ├── llm.service.ts           # Azure OpenAI integration
│   ├── diff-processor.ts        # Diff parsing
│   └── issue-verifier.service.ts # False positive filter
├── queue/
│   └── review-processor.ts      # Bull queue job handler
├── gitlab/
│   └── gitlab.service.ts        # GitLab API client
└── prisma/
    └── prisma.service.ts        # Database client

Database Schema

model Project {
  id              String   @id @default(cuid())
  gitlabProjectId Int      @unique
  name            String
  webhookUrl      String?
  reviews         Review[]
}

model Developer {
  id           String   @id @default(cuid())
  gitlabUserId Int      @unique
  username     String
  email        String?
  reviews      Review[]
}

model Review {
  id              String       @id @default(cuid())
  mergeRequestIid Int
  projectId       String
  developerId     String
  reviewContent   Json
  qualityScore    Int
  issuesFound     Int
  status          ReviewStatus
  createdAt       DateTime     @default(now())
  project         Project      @relation(...)
  developer       Developer    @relation(...)
}

Running Tests

# Backend tests
cd reviewbot-backend
npm test

# Frontend tests
cd reviewbot-frontend
npm run test

Building for Production

# Backend
cd reviewbot-backend
npm run build

# Frontend
cd reviewbot-frontend
npm run build

🐳 Docker Deployment

Option 1: Docker Compose (Full Stack)

cd reviewbot-backend

# Build and start all services
docker-compose up -d

# View logs
docker-compose logs -f

# Stop services
docker-compose down

Services:

PostgreSQL: localhost:5432
Redis: localhost:6379
Backend: localhost:3000
Frontend: localhost:5173

Option 2: Separate Services

# Start only infrastructure
docker-compose up -d postgres redis

# Run backend locally
npm run start:dev

# Run frontend locally
cd ../reviewbot-frontend
npm run dev

🔍 Monitoring & Debugging

Logs

# Backend logs (development)
npm run start:dev

# Docker logs
docker-compose logs -f backend

# Redis queue monitoring
npm run prisma:studio
# Navigate to Review model to see job status

Common Issues

Inline comments not posting:

Verify GITLAB_ACCESS_TOKEN has api scope
Check baseSha, headSha, startSha are valid
Ensure line number is within changed lines in diff

Queue jobs stuck:

Check Redis connection: redis-cli ping
Verify REDIS_HOST and REDIS_PORT in .env
Check Bull queue in Prisma Studio

LLM timeout:

Verify Azure OpenAI credentials
Check deployment name matches AZURE_OPENAI_DEPLOYMENT
Increase timeout in llm.service.ts if needed

🛠️ Advanced Configuration

Batching Threshold

Edit reviewbot-backend/src/queue/review-processor.ts:

// Line 139: Change threshold for batching
const shouldBatch = totalChangedLines <= 500 && allChunksWithContext.length > 1;
//                                       ^^^
//                                       Adjust this value

Context Lines

Edit reviewbot-backend/src/queue/review-processor.ts:

// Line 95: Adjust context lines before/after changes
const chunks = this.diffProcessor.extractChangedLinesWithContext(
  diff.diff,
  10, // Change this value (default: 10)
);

File Limit

Edit reviewbot-backend/src/queue/review-processor.ts:

// Line 19: Maximum files per review
private readonly MAX_FILES = 50; // Adjust to prevent token overflow

📝 API Documentation

Swagger/OpenAPI

Access interactive API documentation at:

http://localhost:3000/api/docs

🤝 Contributing

Contributions welcome! Please follow these steps:

Fork the repository
Create feature branch: git checkout -b feature/amazing-feature
Commit changes: git commit -m 'Add amazing feature'
Push to branch: git push origin feature/amazing-feature
Open Pull Request

📄 License

This project is licensed under the ISC License.

🙏 Acknowledgments

Azure OpenAI: Haiku 4.5 model for fast code analysis
GitLab: Webhook integration and API
NestJS: Robust backend framework
Prisma: Type-safe database ORM
shadcn/ui: Beautiful UI components

📧 Support

For issues, questions, or contributions:

Open an issue on GitHub
Check existing documentation in /docs
Review /plans for project roadmap

Made with ❤️ for better code quality