[build-system]
requires = ["setuptools>=68.0", "wheel"]
build-backend = "setuptools.build_meta"

[project]
name = "nadirclaw"
dynamic = ["version"]
description = "Source-available LLM router: simple prompts to free models, complex to premium"
readme = "README.md"
requires-python = ">=3.10"
license = "PolyForm-Noncommercial-1.0.0"
authors = [{name = "Nadir", email = "nadir@nadirclaw.com"}]
keywords = ["llm", "router", "ai", "openai", "gemini", "anthropic", "cost-optimization", "model-routing"]
classifiers = [
    "Development Status :: 4 - Beta",
    "Intended Audience :: Developers",
    "Programming Language :: Python :: 3",
    "Programming Language :: Python :: 3.10",
    "Programming Language :: Python :: 3.11",
    "Programming Language :: Python :: 3.12",
    "Topic :: Scientific/Engineering :: Artificial Intelligence",
    "Topic :: Software Development :: Libraries :: Python Modules",
]
dependencies = [
    "fastapi>=0.100.0",
    "uvicorn>=0.20.0",
    "litellm>=1.0.0; python_version >= '3.11'",
    # LiteLLM 1.97.0 leaves Message reasoning types unresolved on Python 3.10;
    # 1.98.0 also imports typing.NotRequired (3.11+). Keep 3.10 on the tested
    # 1.96.x line until upstream restores compatibility.
    "litellm>=1.0.0,<1.97.0; python_version < '3.11'",
    "sentence-transformers>=2.0.0",
    "numpy",
    "python-dotenv",
    "click",
    "google-genai>=1.0.0",
    "sse-starlette>=1.0.0",
]

[project.urls]
Homepage = "https://github.com/doramirdor/NadirClaw"
Repository = "https://github.com/doramirdor/NadirClaw"
Issues = "https://github.com/doramirdor/NadirClaw/issues"

[project.scripts]
nadirclaw = "nadirclaw.cli:main"

[tool.setuptools.packages.find]
include = ["nadirclaw*"]

[tool.setuptools.dynamic]
version = {attr = "nadirclaw.__version__"}

[tool.setuptools.package-data]
nadirclaw = [
    "*.npy",
    "cascade_rules/profiles/*.yaml",
    "models/*.pt",
    "tier_config/profiles/*.yaml",
]

[project.optional-dependencies]
dev = [
    "pytest>=7.0",
    "pytest-asyncio>=0.21",
    "httpx",
    # Recent LiteLLM imports its MCP adapter when serializing function tools.
    "mcp>=1.0",
    "pyyaml>=6.0",
]
dashboard = [
    "rich>=13.0",
]
cascade-rules = [
    # PyYAML is only needed to LOAD cascade rule profiles from disk
    # (`load_profile`). `load_inline` works without it, so this is opt-in.
    "pyyaml>=6.0",
]
headroom = [
    # Optional context-compression backend (Apache-2.0, github.com/chopratejas/headroom).
    # Activated with NADIRCLAW_OPTIMIZE_BACKEND=headroom. When absent, the optimizer
    # transparently falls back to the built-in native pipeline, so this stays opt-in.
    # Prebuilt wheels bundle the compiled Rust `_core` extension (SmartCrusher etc.).
    "headroom-ai>=0.23.0",
]
telemetry = [
    "opentelemetry-api>=1.20.0",
    "opentelemetry-sdk>=1.20.0",
    "opentelemetry-exporter-otlp-proto-grpc>=1.20.0",
    "opentelemetry-instrumentation-fastapi>=0.41b0",
]
trained = [
    # Optional dependency for the TrainedVerifier (DeBERTa-v3-small
    # cross-encoder loaded from nadirclaw/cascade-verifier-v1).
    # The heuristic verifier remains the default for the cascade,
    # so users who do not want the transformer stack pay nothing.
    "transformers>=4.40",
    "torch>=2.0",
]

[tool.pytest.ini_options]
testpaths = ["tests"]
asyncio_mode = "auto"