pyproject.toml

[project]
name = "akd"
version = "0.2.0"
description = "Human-centric Multi-Agent System (MAS) framework for scientific discovery"
readme = "README.md"
license = "Apache-2.0"
authors = [
    {name = "Nish", email = "np0069@uah.edu"},
    {name = "Kumar", email = "mr0051@uah.edu"}
]
keywords = ["scientific-discovery", "multi-agent", "research", "literature-search", "ai", "accelerated-knowledge-discovery"]
classifiers = [
    "Development Status :: 3 - Alpha",
    "Intended Audience :: Science/Research",
    "Programming Language :: Python :: 3",
    "Programming Language :: Python :: 3.12",
    "Topic :: Scientific/Engineering",
    "Topic :: Scientific/Engineering :: Information Analysis",
]
requires-python = ">=3.12,<4.0"
# Lightweight core: schemas, protocols, base agent/tool machinery, guardrail
# contracts, mapping. NO ML / browser / LLM-SDK weight here — those live in the
# extras below and are lazy-imported at their call sites (see the lazy-loading
# regression tests in tests/base/test_lazy_imports.py). Install `akd[all]` to
# restore the previous batteries-included behaviour.
dependencies = [
    "loguru>=0.7.3",
    "pydantic>=2.10.3",
    "pydantic-settings>=2.9.1",
    "httpx>=0.28.1",
    "rapidfuzz>=3.13.0",
    "jsonpath-ng>=1.6.1",
    "dateparser>=1.2.2",
]

[project.urls]
Homepage = "https://github.com/NASA-IMPACT/accelerated-discovery"
Repository = "https://github.com/NASA-IMPACT/accelerated-discovery"
Documentation = "https://github.com/NASA-IMPACT/accelerated-discovery/blob/main/README.md"
Issues = "https://github.com/NASA-IMPACT/accelerated-discovery/issues"

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

[tool.setuptools.packages.find]
include = ["akd*"]
exclude = ["tests*", "examples*", "scripts*", "notebooks*"]

[tool.setuptools.package-data]
"akd.guardrails.categories" = ["*.yaml"]

# Pytest configuration
[tool.pytest.ini_options]
asyncio_mode = "auto"
testpaths = ["tests"]
python_files = ["test_*.py", "*_test.py"]
python_classes = ["Test*"]
python_functions = ["test_*"]
addopts = [
    "-v",
    "-n",
    "auto",
    "--dist=loadfile",
    "--strict-markers",
    "--strict-config",
    "-ra",
    "--cov=akd",
    "--cov-report=term-missing",
    "--cov-report=xml",
]
markers = [
    "unit: Unit tests",
    "integration: Integration tests",
    "slow: Slow tests",
    "requires_ml: Tests requiring ML dependencies (sentence-transformers, etc.)",
]

# Optional dependencies (extras)
#
# Layering (each extra sits on top of the lightweight core; extras compose via
# self-references so `pip install akd[risk_agent]` pulls agents + guardrails too):
#
#   core (default)   pydantic · loguru · httpx · rapidfuzz · jsonpath-ng · dateparser
#     ├─ guardrails  + pyyaml            schemas · protocol · risk categories · composite
#     │    ├─ gg     (= guardrails)      GraniteGuardianTool (httpx only)
#     │    └─ risk_agent = agents + deepeval   RiskAgent (deepeval DAG metrics)
#     ├─ agents      + litellm · instructor · openai · tiktoken
#     ├─ scrapers    + crawl4ai · playwright · pymupdf · ...   web/PDF scrapers, resolvers
#     │    └─ omni   + docling · docling-core   DoclingScraper (torch/transformers)
#     ├─ search      + numpy · bs4 · requests · pyyaml   search tools + resolvers
#     └─ ml          + sentence-transformers · pandas · numpy   rerankers, embeddings
#
# `akd[all]` = every runtime extra (previous batteries-included behaviour).
[project.optional-dependencies]
# ── Guardrails ──────────────────────────────────────────────────────────
# Contracts only: GuardrailInput/Output/Protocol + RiskCategory enums (loaded
# from YAML). Lets a downstream service (e.g. akd-guardrails-service) code
# against the lightweight schemas without any ML/LLM weight.
guardrails = [
    "pyyaml>=6.0",
]
# Granite Guardian tool — talks to a served model over HTTP (httpx is in core).
gg = [
    "akd[guardrails]",
]
# Risk agent — deepeval DAG metrics + an LLM agent backend.
risk_agent = [
    "akd[guardrails]",
    "akd[agents]",
    # Cap below 4.0: deepeval 4.x reworked the DAG metric API (DAGMetric/TaskNode/
    # VerdictNode/DeepAcyclicGraph), which RiskAgent builds against. Unbounded,
    # a fresh install pulls 4.x and RiskAgent breaks. Lift once ported to 4.x.
    "deepeval>=3.4.0,<4",
]

# ── Agents (LLM stack) ──────────────────────────────────────────────────
agents = [
    "litellm>=1.81.0",
    "instructor==1.13.0",
    "openai>=1.75.0",
    "tiktoken>=0.9.0",
    # litellm needs orjson on its runtime path but only declares it under its
    # own [proxy] extra, so every akd[agents] consumer must pull it explicitly.
    "orjson>=3.11.6",
]

# ── Scrapers / search ───────────────────────────────────────────────────
scrapers = [
    "crawl4ai>=0.7.6",
    "playwright>=1.40.0",
    "pymupdf>=1.25.4",
    "markdownify>=1.1.0",
    "readability-lxml>=0.8.1",
    "beautifulsoup4>=4.12.0",
    "aiohttp>=3.9.0",
    "requests>=2.32.3",
    "pyyaml>=6.0",
    "gdown>=5.2.0",
    "pypaperbot @ git+https://github.com/NISH1001/PyPaperBot.git@develop",
]
# Docling / omni scraper — pulls the docling (torch/transformers) tree.
omni = [
    "akd[scrapers]",
    "docling>=2.37.0",
    "docling-core>=2.0.0",
]
search = [
    "numpy>=1.26.0",
    "beautifulsoup4>=4.12.0",
    "requests>=2.32.3",
    "pyyaml>=6.0",
]

# ── ML (rerankers / embeddings) ─────────────────────────────────────────
ml = [
    "numpy>=1.26.0",
    "sentence-transformers>=5.0.0",
    "pandas>=2.3.1",
]

# Install this extra if you plan to use AKDSerializer as a langgraph
# checkpoint serde (e.g. `AsyncPostgresSaver(serde=AKDSerializer())`).
serializer = [
    "langgraph==1.0.3",
    "numpy>=1.26.0",
]

# Everything above — restores the previous batteries-included install.
all = [
    "akd[gg]",
    "akd[risk_agent]",
    "akd[omni]",
    "akd[search]",
    "akd[ml]",
    "akd[serializer]",
]

# ── Tooling (not part of `all`) ─────────────────────────────────────────
dev = [
    "pytest>=8.0.0",
    "pytest-asyncio>=1.0.0",
    "pytest-cov>=6.0.0",
    "pytest-xdist>=3.5.0",
    "pre-commit>=4.2.0",
    "memray>=1.18.0",
    "scalene>=1.5.55",
    # litellm imports its proxy pre-call utils (fastapi/starlette) on some
    # completion/health-check paths exercised by the real-LLM integration tests.
    # Not a runtime dep of akd itself — only needed to run the suite.
    "fastapi>=0.110.0",
]
local = [
    "marimo==0.15.0",
    "jupyter",
    "ipykernel>=6.30.0",
    "ipywidgets>=8.1.7",
]

[tool.ruff.lint.per-file-ignores]
"tests/**/*.py" = ["F841"]

[tool.ruff]
target-version = "py312"
line-length = 120

[tool.ruff.format]
# Ensure LF line endings across platforms; used by Ruff formatter on save.
line-ending = "lf"
docstring-code-line-length = 120
quote-style = "double"