pyproject.toml

[build-system]
requires = ["setuptools>=61.0", "wheel", "setuptools-scm>=6.2"]
build-backend = "setuptools.build_meta"

[project]
name = "telefuser"
dynamic = ["version"] 
description = "A framework for efficient model inference with multimodal generation models"
readme = "README.md"
requires-python = ">=3.10,<3.14"
license = {text = "Apache-2.0"}
authors = [
    {name = "TeleAI Team"}
]
keywords = ["multimodal", "diffusion", "transformer", "inference", "serving"]
classifiers = [
    "Development Status :: 3 - Alpha",
    "Intended Audience :: Developers",
    "Intended Audience :: Science/Research",
    "License :: OSI Approved :: Apache Software License",
    "Programming Language :: Python :: 3.10",
    "Programming Language :: Python :: 3.11",
    "Programming Language :: Python :: 3.12",
    "Programming Language :: Python :: 3.13",
    "Topic :: Scientific/Engineering :: Artificial Intelligence",
    "Topic :: Software Development :: Libraries :: Python Modules",
]


dependencies = [
    # Core runtime dependencies (required for actual usage)
    "diffusers>=0.36.0",
    "gradio==5.50",
    "tqdm>=4.66.0",
    "imageio[ffmpeg]>=2.37.2",
    "transformers==4.57.3",
    "safetensors",
    "einops",
    "loguru",
    "openai",
    "fastapi",
    "click", 
    "ftfy",
    "ray",
    "pydantic>=2.0.0",
    "pydantic-settings>=2.0.0",
    "torchvision",
]

[project.optional-dependencies]

# Latent cache subsystem (telefuser.cache_mem).
# Required only when the service is started with enable_latent_cache=true
# (or when running the wan22 service example with cache_mem). Installing
# without this extra leaves the cache layer disabled and adds zero overhead.
cache = [
    "faiss-cpu",
    "scipy",
    "qwen_vl_utils",
    "sentencepiece",
    "protobuf",
]

dev = [
    "pytest>=7.0.0",
    "pytest-asyncio>=0.21.0",
    "pytest-cov>=4.0.0",
    "pre-commit==4.0.1",
    "psutil>=7.2.0",
    "Pillow>=9.0.0",
    "scikit-image>=0.19.0",
    "uv"
]


docs = [
    "mkdocs>=1.5.0",
    "mkdocs-api-autonav",
    "mkdocs-material",
    "mkdocstrings-python",
    "mkdocs-gen-files",
    "mkdocs-awesome-nav",
    "mkdocs-glightbox",
    "mkdocs-git-revision-date-localized-plugin",
    "mkdocs-minify-plugin",
    "mkdocs-static-i18n",
    "regex",
    "ruff==0.9.4",
]

webrtc = [
    "aiortc>=1.9.0",
    "opencv-python-headless>=4.0",
]


[project.urls]
Homepage = "https://www.srdcloud.cn/code/49596/repoView/372413/codeFile/tree"
Repository = "https://www.srdcloud.cn/code/49596/repoView/372413/codeFile/tree"

[project.scripts]
telefuser = "telefuser.entrypoints.cli.main:main"


[tool.setuptools.packages.find]
where = ["."]
include = ["telefuser*"]

[tool.pytest.ini_options]
minversion = "7.0"
testpaths = ["tests"]
python_files = ["test_*.py"]
python_classes = ["Test*"]
python_functions = ["test_*"]
addopts = [
    "-v",
    "--strict-markers",
    "--tb=short",
    "-ra",
]
markers = [
    "slow: marks tests as slow",
    "gpu: marks tests that require GPU",
    "multi_gpu: marks tests that require multiple GPUs",
    "distributed: marks tests that require distributed setup",
    "quant: marks tests that require quantization support",
    "filesystem: marks tests that require filesystem access (skipped in CI)",
    "integration: marks tests that start a real server and make HTTP requests",
]

[tool.coverage.run]
source = ["telefuser"]
omit = [
    "*/tests/*",
    "*/tf-kernel/*",
    "*/examples/*",
    # Modules that require GPU/models - temporarily excluded from coverage
    "*/models/*",
    "*/pipelines/*",
    "*/orchestrator/*",
    "*/kernel/*",
    "*/service/*.py",
    "*/worker/*",
    "*/utils/hf_utils.py",
    "*/utils/lora_loader.py",
    "*/utils/profiler.py",
    "*/utils/system.py",
    "*/utils/torch_compile.py",
    "*/utils/video.py",
    "*/utils/func.py",
    "*/distributed/a2a.py",
    "*/distributed/fsdp.py",
    "*/distributed/parallel_shard.py",
    "*/distributed/tp_parallelize.py",
    "*/schedulers/lcm.py",
    "*/schedulers/unipc.py",
    "*/feature_cache/ada_taylor_cache/params/*",
]
branch = true

[tool.coverage.report]
exclude_lines = [
    "pragma: no cover",
    "def __repr__",
    "raise AssertionError",
    "raise NotImplementedError",
    "if __name__ == .__main__.:",
    "@pytest.mark.gpu",
]
# fail_under = 40  # Temporarily disabled until more tests are added

[tool.ruff]
line-length = 120
exclude = [
    ".eggs",
    ".git",
    ".hg",
    ".mypy_cache",
    ".tox",
    ".venv",
    "build",
    "dist",
    "tf-kernel",
    "benchmarks",
    "telefuser/_version.py"
]

[tool.ruff.lint]
select = [
    "E",   # pycodestyle errors
    "W",   # pycodestyle warnings
    "F",   # pyflakes
    "I",   # isort
    "ANN", # flake8-annotations: type annotation checks
]
ignore = [
    "E203",   # whitespace before ':' (conflicts with black)
    "F401",
    "E402",
    "E712",
    # Type annotation rules - adjust based on project needs
    "ANN401", # Allow dynamically typed expressions (Any) for flexibility
    # Return type annotations - disabled to reduce annotation burden
    "ANN001", # Missing return type annotation for public function
    "ANN201", # Missing return type annotation for public function
    "ANN202", # Missing return type annotation for private method
    "ANN204", # Missing return type annotation for special method (__init__)
    "ANN205", # Missing return type annotation for static method
    "ANN206", # Missing return type annotation for class method
    # Variable argument annotations - disabled for flexibility
    "ANN002", # Missing type annotation for `*args`
    "ANN003", # Missing type annotation for `**kwargs`
]

[tool.ruff.lint.per-file-ignores]
"examples/**" = ["E501", "ANN"]  # Allow long lines and missing annotations in examples
"tests/**" = ["E501", "ANN"]     # Allow long lines and missing annotations in tests
"telefuser/kernel/triton/**" = ["ANN"]  # Allow missing annotations in triton kernels
[tool.ruff.lint.isort]
known-first-party = ["telefuser"]
case-sensitive = true
[tool.setuptools_scm]
write_to = "telefuser/_version.py"