[project]
name = "lyxy-document"
version = "0.1.0"
description = "帮助AI工具读取转换文档到markdown的skill"
readme = "README.md"
requires-python = ">=3.11"
dependencies = [
    "chardet>=5.0.0",
]

[project.optional-dependencies]
docx = [
    "docling>=2.0.0",
    "unstructured>=0.12.0",
    "markitdown>=0.1.0",
    "pypandoc-binary>=1.13.0",
    "python-docx>=1.1.0",
    "markdownify>=0.12.0",
]
xlsx = [
    "docling>=2.0.0",
    "unstructured>=0.12.0",
    "markitdown>=0.1.0",
    "pandas>=2.0.0",
    "tabulate>=0.9.0",
]
pptx = [
    "docling>=2.0.0",
    "unstructured>=0.12.0",
    "markitdown>=0.1.0",
    "python-pptx>=0.6.0",
    "markdownify>=0.12.0",
]
pdf = [
    "docling>=2.0.0",
    "unstructured>=0.12.0",
    "unstructured-paddleocr>=0.1.0",
    "markitdown>=0.1.0",
    "pypdf>=4.0.0",
    "markdownify>=0.12.0",
]
html = [
    "trafilatura>=1.10.0",
    "domscribe>=0.1.0",
    "markitdown>=0.1.0",
    "html2text>=2024.2.26",
    "beautifulsoup4>=4.12.0",
]
http = [
    "httpx>=0.27.0",
    "pyppeteer>=2.0.0",
    "selenium>=4.18.0",
]
office = [
    "lyxy-document[docx,xlsx,pptx,pdf]",
]
web = [
    "lyxy-document[html,http]",
]
full = [
    "lyxy-document[office,web]",
]
dev = [
    "pytest>=8.0.0",
    "pytest-cov>=4.1.0",
]