ai-api-wrapper 0.1.3

A unified API wrapper for various AI providers with proxy support

AI API Wrapper

A unified API wrapper for various AI providers, with built-in proxy support.

Features

• 🔄 Unified Interface: Access different AI models through a single, consistent API
• 🌐 Proxy Support: Built-in proxy support, perfect for users in mainland China
• 🤖 Multiple Providers: Support for various AI providers including:
  • OpenAI
  • Anthropic
  • Azure OpenAI
  • Google AI
  • DeepSeek
  • More providers coming soon...

Installation

# Using poetry
poetry install

# Or using pip
pip install ai-api-wrapper

Quick Start

from ai_api_wrapper import Client

# Create a client with proxy support
proxy_config = {
    "http": "http://127.0.0.1:7890",
    "https": "http://127.0.0.1:7890"
}
client = Client(proxy_config=proxy_config)

# Create a chat completion
response = client.chat.completions.create(
    model="deepseek:deepseek-chat",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Hello!"}
    ],
    temperature=0.7
)

# Print the response
print(response.choices[0].message.content)

Configuration

Environment Variables

Create a .env file in your project root with the following variables:

# API Keys
DEEPSEEK_API_KEY=your_api_key_here
OPENAI_API_KEY=your_api_key_here
ANTHROPIC_API_KEY=your_api_key_here
GOOGLE_API_KEY=your_api_key_here
AZURE_OPENAI_API_KEY=your_api_key_here

# Base URLs (Optional, will override config.json if set)
DEEPSEEK_BASE_URL=https://api.deepseek.com
OPENAI_BASE_URL=https://api.openai.com
ANTHROPIC_BASE_URL=https://api.anthropic.com
GOOGLE_BASE_URL=https://generativelanguage.googleapis.com
AZURE_OPENAI_BASE_URL=https://your-resource.openai.azure.com

Configuration File

The project uses a single config.json file for non-sensitive configurations. Base URLs in this file will be overridden by environment variables if set:

{
    "deepseek": {
        "base_url": "https://api.deepseek.com",
        "timeout": 30.0,
        "max_retries": 3,
        "verify_ssl": true,
        "models": {
            "deepseek-chat": {
                "max_tokens": 4000,
                "temperature": 0.7
            },
            "deepseek-coder": {
                "max_tokens": 4000,
                "temperature": 0.7
            }
        }
    },
    "openai": {
        "base_url": "https://api.openai.com",
        "timeout": 30.0,
        "max_retries": 3,
        "verify_ssl": true
    },
    "anthropic": {
        "base_url": "https://api.anthropic.com",
        "timeout": 30.0,
        "max_retries": 3,
        "verify_ssl": true
    },
    "google": {
        "base_url": "https://generativelanguage.googleapis.com",
        "timeout": 30.0,
        "max_retries": 3,
        "verify_ssl": true
    },
    "azure": {
        "base_url": "https://your-resource.openai.azure.com",
        "timeout": 30.0,
        "max_retries": 3,
        "verify_ssl": true
    },
    "logging": {
        "level": "INFO",
        "file": "app.log"
    }
}

Security Best Practices

1. Never commit sensitive information

   • Keep all API keys in .env file
   • Add .env to your .gitignore
   • Never commit API keys to version control

2. Configuration file

   • Keep config.json in version control
   • Only include non-sensitive configurations
   • Use environment variables for sensitive data and custom base URLs

3. Environment variables

   • Use .env file for local development
   • Use system environment variables in production
   • Never hardcode API keys or base URLs in your code

Development

# Install development dependencies
poetry install --with dev

# Run tests
poetry run pytest

# Format code
poetry run black .
poetry run isort .

# Type checking
poetry run mypy .

License

GPL v3

配置系统说明

配置文件结构

项目使用分层的JSON配置文件系统：

1. default.json：默认配置文件

   • 包含所有可能的配置项
   • 定义默认值和基本结构
   • 可以提交到版本控制
   • 不应包含任何敏感信息（API密钥、密码等）

2. local.json：本地配置文件

   • 包含敏感信息（API密钥、密码等）
   • 覆盖默认配置
   • 不应提交到版本控制
   • 应添加到 .gitignore 文件中

安全最佳实践

1. 敏感信息仅存储在 local.json 中

   • API 密钥
   • 数据库凭据
   • JWT 密钥
   • 任何其他密码或密钥

2. 禁止在 default.json 中存储任何真实的敏感信息

   • 在 default.json 中，将所有密钥字段设置为空字符串 ""
   • 仅使用 default.json 定义配置结构

3. 版本控制注意事项

   • 始终将 local.json 添加到 .gitignore
   • 仅提交 default.json 到仓库
   • 提供 local.json.template 文件，用于新开发人员设置

配置项说明

AI服务配置

{
    "ai_services": {
        "service_name": {
            "enabled": true,
            "default_provider": "official",
            "default_model": "model-name",
            "providers": {
                "provider_name": {
                    "name": "显示名称",
                    "enabled": true,
                    "api_key": "",  // 在 default.json 中留空，在 local.json 中填写实际值
                    "base_url": "API基础URL",
                    "models": {
                        "model_name": {
                            "internal_name": "提供商特定的模型名称",
                            "max_tokens": 4000,
                            "temperature": 0.7
                        }
                    }
                }
            }
        }
    }
}

其他配置

• database：数据库配置（密码仅存储在 local.json）
• redis：Redis配置（密码仅存储在 local.json）
• jwt_secret：JWT密钥（仅存储在 local.json）
• logging：日志配置

使用方法

1. 复制local.json.template为local.json
2. 在local.json中填入实际的配置值（API密钥等）
3. 不要修改default.json中的默认值（除非添加新功能）
4. 使用ConfigManager类访问配置：

from utils.config_manager import ConfigManager

config = ConfigManager()

# 获取提供商配置
provider_config = config.get_provider_config('deepseek', 'siliconflow')

# 获取模型配置
model_config = config.get_model_config('deepseek', 'siliconflow', 'deepseek-chat')

# 获取已启用的服务
enabled_services = config.get_enabled_services()

Test

python -m pytest tests/ -v

注意事项

1. 不要在代码中硬编码任何敏感信息
2. 不要提交local.json到版本控制
3. 保持配置文件的结构与default.json一致
4. 使用有意义的配置项名称
5. 及时更新文档