# Telegram ChatBot with Ollama AI

A high-quality, production-ready Telegram chatbot powered by Ollama AI models. This bot provides natural conversation experiences using local AI models.

## 🎯 Features

- **Ollama Integration**: Uses OllamaSharp library for efficient AI model communication
- **Multiple Model Support**: Automatically manages and switches between multiple AI models
- **Session Management**: Maintains conversation history for each chat
- **Command System**: Extensible command architecture for bot commands
- **Smart Retry Logic**: Exponential backoff with jitter for failed requests
- **Rate Limit Handling**: Automatic model switching on rate limits
- **Natural Conversation**: Configurable response delays for human-like interactions
- **Group Chat Support**: Works in both private and group conversations
- **Robust Logging**: Comprehensive logging with Serilog

## 📋 Prerequisites

- .NET 9.0 or later
- Ollama server running locally or remotely
- Telegram Bot Token (from [@BotFather](https://t.me/botfather))

## 🚀 Getting Started

### 1. Install Ollama

Download and install Ollama from [ollama.ai](https://ollama.ai)

### 2. Pull an AI Model

```bash
ollama pull llama3
```

### 3. Configure the Bot

Edit `appsettings.json`:

```json
{
  "TelegramBot": {
    "BotToken": "YOUR_BOT_TOKEN_HERE"
  },
  "Ollama": {
    "Url": "http://localhost:11434",
    "MaxRetries": 3,
    "MaxTokens": 1000,
    "Temperature": 0.7,
    "ResponseDelay": {
      "IsEnabled": true,
      "MinDelayMs": 1000,
      "MaxDelayMs": 3000
    },
    "SystemPromptFilePath": "system-prompt.txt"
  }
}
```

Edit `appsettings.Models.json` to configure your models:

```json
{
  "ModelConfigurations": [
    {
      "Name": "llama3",
      "MaxTokens": 2000,
      "Temperature": 0.8,
      "Description": "Llama 3 Model",
      "IsEnabled": true
    }
  ]
}
```

### 4. Customize System Prompt

Edit `system-prompt.txt` to define your bot's personality and behavior.

### 5. Run the Bot

```bash
cd ChatBot
dotnet run
```

## 🏗️ Architecture

### Core Services

- **AIService**: Handles AI model communication and text generation
- **ChatService**: Manages chat sessions and message history
- **ModelService**: Handles model selection and switching
- **TelegramBotService**: Main Telegram bot service

### Command System

Commands are automatically registered using attributes:

```csharp
[Command("start", "Start conversation with the bot")]
public class StartCommand : TelegramCommandBase
{
    // Implementation
}
```

Available commands:
- `/start` - Start conversation
- `/help` - Show help information
- `/clear` - Clear conversation history
- `/settings` - View current settings

## ⚙️ Configuration

### Ollama Settings

- **Url**: Ollama server URL
- **MaxRetries**: Maximum retry attempts for failed requests
- **MaxTokens**: Default maximum tokens for responses
- **Temperature**: AI creativity level (0.0 - 2.0)
- **ResponseDelay**: Add human-like delays before responses
- **SystemPromptFilePath**: Path to system prompt file

### Model Configuration

Each model can have custom settings:

- **Name**: Model name (must match Ollama model name)
- **MaxTokens**: Maximum tokens for this model
- **Temperature**: Temperature for this model
- **Description**: Human-readable description
- **IsEnabled**: Whether the model is available for use

## 🔧 Advanced Features

### Automatic Model Switching

The bot automatically switches to alternative models when:
- Rate limits are encountered
- Current model becomes unavailable

### Session Management

- Automatic session creation per chat
- Configurable message history length
- Old session cleanup (default: 24 hours)

### Error Handling

- Exponential backoff with jitter for retries
- Graceful degradation on failures
- Comprehensive error logging

## 📝 Development

### Project Structure

```
ChatBot/
├── Models/
│   ├── Configuration/      # Configuration models
│   │   └── Validators/     # Configuration validation
│   └── Dto/                # Data transfer objects
├── Services/
│   ├── Telegram/           # Telegram-specific services
│   │   ├── Commands/       # Bot commands
│   │   ├── Interfaces/     # Service interfaces
│   │   └── Services/       # Service implementations
│   ├── AIService.cs        # AI model communication
│   ├── ChatService.cs      # Chat session management
│   └── ModelService.cs     # Model management
└── Program.cs              # Application entry point
```

### Adding New Commands

1. Create a new class in `Services/Telegram/Commands/`
2. Inherit from `TelegramCommandBase`
3. Add `[Command]` attribute
4. Implement `ExecuteAsync` method

Example:

```csharp
[Command("mycommand", "Description of my command")]
public class MyCommand : TelegramCommandBase
{
    public override async Task ExecuteAsync(TelegramCommandContext context)
    {
        await context.MessageSender.SendTextMessageAsync(
            context.Message.Chat.Id,
            "Command executed!"
        );
    }
}
```

## 🐛 Troubleshooting

### Bot doesn't respond

1. Check if Ollama server is running: `ollama list`
2. Verify bot token in `appsettings.json`
3. Check logs in `logs/` directory

### Model not found

1. Pull the model: `ollama pull model-name`
2. Verify model name matches in `appsettings.Models.json`
3. Check model availability: `ollama list`

### Connection errors

1. Verify Ollama URL in configuration
2. Check firewall settings
3. Ensure Ollama server is accessible

## 📦 Dependencies

- **OllamaSharp** (v5.4.7): Ollama API client
- **Telegram.Bot** (v22.7.2): Telegram Bot API
- **Serilog** (v4.3.0): Structured logging
- **Microsoft.Extensions.Hosting** (v9.0.10): Host infrastructure

## 📄 License

This project is licensed under the terms specified in [LICENSE.txt](LICENSE.txt).

## 🤝 Contributing

Contributions are welcome! Please ensure:
- Code follows existing patterns
- All tests pass
- Documentation is updated
- Commits are descriptive

## 🔮 Future Enhancements

- [ ] Multi-language support
- [ ] Voice message handling
- [ ] Image generation support
- [ ] User preferences persistence
- [ ] Advanced conversation analytics
- [ ] Custom model fine-tuning support

---

Built with ❤️ using .NET 9.0 and Ollama