pyproject.toml

[build-system]
requires = ["setuptools>=61.0", "wheel"]
build-backend = "setuptools.build_meta"

[project]
name = "hello-agents"
version = "0.2.8"
description = "灵活、可扩展的多智能体框架 - 基于OpenAI原生API"
readme = "README.md"
license = {text = "CC-BY-NC-SA-4.0"}
authors = [
    {name = "HelloAgents Team", email = "jjyaoao@126.com"}
]
requires-python = ">=3.10"
dependencies = [
    # 核心依赖
    "openai>=1.0.0,<2.0.0",
    "requests>=2.25.0,<3.0.0",
    "python-dotenv>=0.19.0,<2.0.0",
    "pydantic>=2.0.0,<3.0.0",

    # 基础工具库
    "beautifulsoup4>=4.9.0,<5.0.0",
    "numpy>=2.0.0,<3.0.0",
    "networkx>=2.6.0,<4.0.0",
    "tiktoken>=0.5.0",

]

[project.optional-dependencies]
# 搜索功能依赖
search = [
    "tavily-python>=0.7.12",
    "google-search-results>=2.4.2",
]

# 记忆系统依赖
memory = [
    "qdrant-client>=1.9.0",          # 向量数据库（已兼容1.16.0+新API）
    "neo4j>=5.0.0",                  # 图数据库
    "spacy>=3.4.0",                  # NLP和实体识别
    "scikit-learn>=1.0.0",           # 机器学习工具
]

# RAG系统依赖
rag = [
    "scikit-learn>=1.0.0",       # TF-IDF支持
    "transformers>=4.20.0",      # Transformer模型
    "torch>=1.12.0",             # PyTorch
    "sentence-transformers>=2.2.0",  # 句子嵌入
    "markitdown>=0.0.1",         # 文档转换
    "pypdf>=3.9.0",              # PDF处理
    "pdfminer.six>=20221105",    # PDF解析备选
]

# 协议支持依赖
protocols = [
    "fastmcp>=2.0.0,<3.0.0",        # MCP 协议支持
    "a2a-sdk>=0.1.0",               # A2A 协议支持（可选）
]

# 评估系统依赖（不含BFCL）
evaluation = [
    "datasets>=2.14.0",             # HuggingFace数据集加载
    "huggingface_hub>=0.20.0,<1.0.0",  # HuggingFace Hub API
    "evaluate>=0.4.0,<1.0.0",       # 评估工具
    "pandas>=2.0.0,<3.0.0",         # 数据处理和分析
    "matplotlib>=3.7.0,<4.0.0",     # 可视化(可选)
    "seaborn>=0.12.0,<1.0.0",       # 高级可视化(可选)
    "tqdm>=4.65.0,<5.0.0",          # 进度条显示
    "gradio>=4.0.0,<5.0.0",         # 人工验证界面
]

# BFCL评估依赖（⚠️ bfcl-eval强制要求numpy==1.26.4，与主依赖numpy>=2.0冲突）
# ⚠️ 此组已禁用，无法通过 pip install hello-agents[evaluation-bfcl] 安装
# 
# 使用方法：在独立虚拟环境中手动安装
#   uv venv .venv-bfcl
#   source .venv-bfcl/bin/activate  # Windows: .venv-bfcl\Scripts\activate
#   uv pip install openai requests python-dotenv pydantic beautifulsoup4 networkx
#   uv pip install "numpy==1.26.4" bfcl-eval datasets pandas tqdm huggingface_hub evaluate
#
# evaluation-bfcl = [
#     "datasets>=2.14.0",
#     "huggingface_hub>=0.20.0,<1.0.0",
#     "evaluate>=0.4.0,<1.0.0",
#     "pandas>=2.0.0,<3.0.0",
#     "tqdm>=4.65.0,<5.0.0",
#     "bfcl-eval",                    # BFCL官方评估工具（强制numpy==1.26.4）
# ]

# RL训练系统依赖（第11章）
rl = [
    "trl>=0.24.0",                  # Transformer Reinforcement Learning
    "transformers>=4.20.0",         # Transformer模型
    "torch>=2.0.0",                 # PyTorch
    "datasets>=2.14.0",             # 数据集加载（兼容trl）
    "accelerate>=0.20.0",           # 分布式训练加速
    "peft>=0.5.0",                  # LoRA等参数高效微调
    "bitsandbytes>=0.41.0",         # 量化支持
    "wandb>=0.15.0",                # 训练监控（可选）
    "tensorboard>=2.13.0",          # TensorBoard支持（可选）
]

# 完整记忆和RAG功能
memory-rag = [
    "hello-agents[memory,rag]",
]

# 全部功能（不含evaluation-bfcl，因其与numpy>=2.0冲突）
all = [
    "hello-agents[search,memory-rag,protocols,evaluation,rl]",
]

[project.urls]
Homepage = "https://github.com/jjyaoao/HelloAgents"
Documentation = "https://github.com/jjyaoao/HelloAgents/blob/main/README.md"
Repository = "https://github.com/jjyaoao/HelloAgents"
"Bug Tracker" = "https://github.com/jjyaoao/HelloAgents/issues"

[tool.setuptools.packages.find]
where = ["."]
include = ["hello_agents*"]
exclude = ["tests*", "examples*", "docs*"]

[tool.setuptools.package-data]
hello_agents = ["py.typed"]

[tool.black]
line-length = 88
target-version = ['py310']
include = '\.pyi?$'
extend-exclude = '''
/(
  # directories
  \.eggs
  | \.git
  | \.hg
  | \.mypy_cache
  | \.tox
  | \.venv
  | build
  | dist
)/
'''

[tool.isort]
profile = "black"
multi_line_output = 3
line_length = 88

[tool.pytest.ini_options]
testpaths = ["tests"]
python_files = ["test_*.py"]
python_classes = ["Test*"]
python_functions = ["test_*"]
addopts = "-v --tb=short"

[tool.mypy]
python_version = "3.10"
warn_return_any = true
warn_unused_configs = true
disallow_untyped_defs = true