pyproject.toml

[build-system]
requires = ["uv_build>=0.8.15,<0.9.0"]
build-backend = "uv_build"

[project]
name = "attachments"
version = "1.0.0a1"  # pre-release: pip install attachments will NOT pick this up; see VISION.md
description = "Turn anything into LLM-ready artifacts. Universal file and source processing."
readme = "README.md"
requires-python = ">=3.12"
authors = [{ name = "Maxime Rivest", email = "mrive052@gmail.com" }]
license = "MIT"
license-files = ["LICENSE"]
keywords = [
    "llm",
    "context",
    "pdf",
    "docx",
    "xlsx",
    "pptx",
    "extraction",
    "rag",
    "prompt",
]
classifiers = [
    "Development Status :: 5 - Production/Stable",
    "Intended Audience :: Developers",
    "Programming Language :: Python :: 3.12",
    "Programming Language :: Python :: 3.13",
    "Topic :: Software Development :: Libraries",
    "Topic :: Text Processing",
]

# ZERO required dependencies - everything is optional
dependencies = []

[project.optional-dependencies]
# === Individual Processors ===
pdf = [
    "pypdf>=4.0",           # Modern pypdf (not pypdf2)
    "pymupdf>=1.24",        # For image rendering
]
pdf-fallback = [
    "pdfminer.six>=20221105",  # Fallback text extraction
]
xlsx = [
    "openpyxl>=3.1,<4.0",
]
xlsx-pandas = [
    "pandas>=2.0,<3.0",
    "openpyxl>=3.1,<4.0",
]
docx = [
    "python-docx>=1.0",
]
pptx = [
    "python-pptx>=0.6.23",
]
html = [
    "beautifulsoup4>=4.12",
    "lxml>=5.0",
]
image = [
    "pillow>=10.0",
]
heic = [
    "pillow>=10.0",
    "pillow-heif>=0.16",
]
svg = [
    "cairosvg>=2.7",
]
xls = [
    "xlrd>=2.0",
]
csv-pandas = [
    "pandas>=2.0,<3.0",
]
ocr = [
    "rapidocr-onnxruntime>=1.3",
]
audio = [
    "faster-whisper>=1.0",
]
mcp = [
    "mcp>=1.2",
]

# === Service Mode (minimal - just HTTP client) ===
service = [
    "httpx>=0.27",
]

# === CLI clipboard support (`att --copy`) ===
clipboard = [
    "pyperclip>=1.8",
]

# === Convenience Bundles ===
# Only formats with shipped processors belong here. New processors (audio,
# s3://, gdrive://, ...) are added as they land - see VISION.md.
office = [
    "attachments[xlsx,docx,pptx]",
]
all-local = [
    "attachments[pdf,pdf-fallback,xlsx-pandas,docx,pptx,html,image,heic,svg,xls,csv-pandas]",
]

# === Self-Hosted Server ===
# Install this on your server machine to serve all formats to clients.
# ocr (onnxruntime) and audio (ctranslate2 via faster-whisper) ship very
# large wheels, so they are deliberately NOT in all-local — local installs
# opt in with attachments[ocr] / attachments[audio]. The server bundle
# includes them because a server should serve everything.
server = [
    "attachments[all-local,ocr,audio]",
]

[project.urls]
Homepage = "https://github.com/MaximeRivest/attachments"
Repository = "https://github.com/MaximeRivest/attachments"
Documentation = "https://github.com/MaximeRivest/attachments"

[dependency-groups]
dev = [
    "ruff",
    "pytest",
    "pytest-cov",
    "pre-commit",
    # Service mode for testing
    "httpx>=0.27",
    # Common processors for testing
    "pypdf>=4.0",
    "pymupdf>=1.24",
    "openpyxl>=3.1,<4.0",
    "pandas>=2.0,<3.0",
    "python-docx>=1.0",
    "python-pptx>=0.6.23",
    "beautifulsoup4>=4.12",
    "lxml>=5.0",
    "pillow>=10.0",
    "pillow-heif>=0.16",
    "cairosvg>=2.7",
    "xlrd>=2.0",
    "rapidocr-onnxruntime>=1.3",
    "faster-whisper>=1.0",
    "mcp>=1.2",
    # CLI clipboard support (`att --copy`)
    "pyperclip>=1.8",
    # Spec conformance validation
    "jsonschema>=4.21",
    # Dev tools
    "ipykernel>=6.30.1",
    "gunicorn>=25.0",
]

[project.scripts]
attachments-server = "attachments.server:main"
attachments-mcp = "attachments.mcp_server:main"
attachments = "attachments.cli:app"
att = "attachments.cli:app"

[tool.ruff]
line-length = 88
target-version = "py312"
extend-exclude = ["docs/_freeze", "build", "dist"]

[tool.ruff.lint]
select = ["E", "F", "I", "B", "UP"]

[tool.ruff.format]
docstring-code-format = true

[tool.pytest.ini_options]
minversion = "6.0"
addopts = "-ra -q --strict-markers --color=yes --cov=attachments --cov-report=term-missing --doctest-modules"
testpaths = ["tests", "src"]