Skip to main content

SAGE Middleware - Streaming-Augmented Generative Execution

Project description

SAGE Middleware(中间件)

用于构建带有 AI 能力的流式数据应用的中间件层

Python Version License PyPI

📋 Overview

SAGE Middleware 是 SAGE 框架的 L4 中间件层,提供高性能的数据处理组件和 AI 能力集成。它集成了多家大模型提供商、异步任务调度、安全鉴权以及核心数据处理组件(向量数据库、流处理引擎等)。

🧭 Governance / 团队协作制度

  • docs/governance/TEAM.md
  • docs/governance/MAINTAINERS.md
  • docs/governance/DEVELOPER_GUIDE.md
  • docs/governance/PR_CHECKLIST.md
  • docs/governance/SELF_HOSTED_RUNNER.md
  • docs/governance/TODO.md

✨ Key Features

  • 🤖 LLM 推理
    • sageLLM ✅ 推荐:统一 LLM 推理引擎,支持 CUDA/Ascend/Mock 后端
    • vLLM ⚠️ 已弃用:将在 v0.4.0 移除,请迁移至 sageLLM
  • 🔎 检索与向量:RAG、BM25、FAISS 等
  • 📋 任务调度:Celery 异步任务
  • 🔐 安全鉴权:JWT、密码学工具
  • ⚙️ 核心组件:
    • sage_db:数据库/向量存储相关组件(含 C/C++ 扩展)
    • sage_flow:高性能向量流处理(可能包含扩展或独立子模块)

🚀 Installation

Basic Installation

pip install isage-middleware

Development Installation

git clone https://github.com/intellistream/SAGE.git
cd SAGE/packages/sage-middleware
pip install -e .

说明:中间件组件(sage_db/sage_flow/sage_tsdb 等)现已随源码直接提供或通过 pip 依赖分发,无需初始化任何子模块。

With Optional Dependencies

# 可选:VLLM 支持(需要 CUDA)
pip install isage-common[vllm]

# 可选:与完整 SAGE 框架集成
pip install isage-middleware[sage]

📖 Quick Start

LLM 推理(推荐:sageLLM)

from sage.middleware.operators.llm import SageLLMGenerator

# 自动选择最佳后端
generator = SageLLMGenerator(
    model_path="Qwen/Qwen2.5-7B-Instruct",
    backend_type="auto",  # auto/cuda/ascend/mock
    temperature=0.7,
    max_tokens=2048,
)

result = generator.execute("你好,世界!")
print(result)

API 客户端

from sage.middleware.api.client import APIClient
from sage.middleware.auth.jwt import JWTManager

client = APIClient()
jwt_manager = JWTManager()

resp = client.chat_completion(
    provider="openai",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(resp)

📖 迁移指南:如果您正在使用 VLLMGenerator,请参阅 vLLM to sageLLM Migration Guide

📦 Package Structure

sage-middleware/
├── src/
│   └── sage/
│       └── middleware/
│           ├── __init__.py
│           ├── operators/          # Dataflow operators
│           │   ├── llm/           # LLM operators (sageLLM)
│           │   ├── rag/           # RAG operators
│           │   └── agentic/       # Agent operators
│           ├── components/         # Core components
│           │   ├── sage_db/       # Vector database (C++ extensions)
│           │   ├── sage_flow/     # Stream processing (C++ extensions)
│           │   ├── sage_tsdb/     # Time-series DB (C++ extensions)
│           │   ├── sage_mem/      # Memory management
│           │   └── sage_refiner/  # Context compression
│           ├── api/                # API clients
│           └── auth/               # Authentication
├── tests/
│   ├── unit/
│   └── integration/
├── docs/
│   └── governance/                # Team collaboration
├── README.md
├── pyproject.toml
└── setup.py

🔧 Configuration

配置可以通过以下方式提供:

  • 环境变量
  • YAML/TOML 配置文件
  • 直接通过 API 参数

配置示例

# config.yaml
middleware:
  auth:
    secret_key: "your-secret-key"  # pragma: allowlist secret
    algorithm: "HS256"
  providers:
    openai:
      api_key: "sk-..."  # pragma: allowlist secret
      base_url: "https://api.openai.com/v1"

📚 Documentation

🧪 Testing

# 运行单元测试
pytest tests/unit -v

# 运行集成测试
pytest tests/integration -v

# 运行所有测试并生成覆盖率报告
pytest --cov=sage.middleware --cov-report=html

🤝 Contributing

欢迎贡献!请参阅:

新增中间件组件规范

当添加新的中间件组件时,请遵循以下要求:

  1. 目录结构: 在 src/sage/middleware/components/ 下创建组件目录
  2. C++ 扩展: 如包含 C++ 扩展,必须遵守依赖约束(参见下文详细规范)
  3. 构建脚本: 提供标准的 build.sh,支持 --install-deps
  4. setup.py 集成: 在 setup.py 中添加构建方法
  5. 测试: 添加单元测试和集成测试
  6. 文档: 更新 README 和相关文档

📄 License

本项目采用 MIT License - 详见 LICENSE 文件

🔗 Related Packages

  • sage-kernel: 核心计算引擎(L3)
  • sage-common: 通用工具(L1)
  • sage-libs: 可复用算法库(L3)
  • sage-platform: 平台服务(L2)
  • isage-vdb: 向量数据库(独立包)
  • isage-neuromem: 内存管理系统(独立包)
  • isage-refiner: 上下文压缩(独立包)

📮 Support


Part of the SAGE Framework | 主仓库

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distributions

No source distribution files available for this release.See tutorial on generating distribution archives.

Built Distributions

If you're not sure about the file name format, learn more about wheel file names.

isage_middleware-0.2.4.13-py3-none-any.whl (2.9 MB view details)

Uploaded Python 3

isage_middleware-0.2.4.13-py2.py3-none-any.whl (1.1 MB view details)

Uploaded Python 2Python 3

File details

Details for the file isage_middleware-0.2.4.13-py3-none-any.whl.

File metadata

File hashes

Hashes for isage_middleware-0.2.4.13-py3-none-any.whl
Algorithm Hash digest
SHA256 30830b5e5bef740c1c3f9e9db061adbe26c58bce4a2e6491bc78c7ef1396d134
MD5 39ae4cb569ca72a00d22dbb1c6944a9b
BLAKE2b-256 d1feb99358e18f05911edf9373a385fdc86483b10255decf4905716b84a15fdc

See more details on using hashes here.

File details

Details for the file isage_middleware-0.2.4.13-py2.py3-none-any.whl.

File metadata

File hashes

Hashes for isage_middleware-0.2.4.13-py2.py3-none-any.whl
Algorithm Hash digest
SHA256 88851a2aab14c37c66b72c25b0f433ddc241524baa0804408cea39e4f3ab9769
MD5 fdd4b5173eaedd917c2031949038eae0
BLAKE2b-256 9890b450fd403cd682f33399eb316bbbe86303e635e203acc2baffbbc7ee43b8

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page