HTML Parse

Parse HTML documents into structured Markdown using MinerU. Analyzes HTML structure and converts it into well-organized Markdown preserving hierarchy and formatting. Features: structural HTML parsing. Preserves headings, lists, tables, and nested elements. Converts HTML to structured Markdown. Works with local files and URLs. Use when you need to: parse HTML structure, convert HTML to structured Markdown, analyze HTML document layout, extract structured content from web pages. Use when asked: 'how do I parse this HTML', 'convert HTML to Markdown with structure', 'I want to understand this HTML layout', 'can my agent parse HTML files', 'is there a skill for HTML parsing'. Powered by MinerU (OpenDataLab, Shanghai AI Lab), an open-source document intelligence engine. Ideal for developers, content managers, and data pipelines that need to parse and restructure HTML content.

Parse local HTML files into structured Markdown using MinerU. Preserves document hierarchy. For live web pages, use mineru-open-api crawl.

Install

npm install -g mineru-open-api
# or via Go (macOS/Linux):
go install github.com/opendatalab/MinerU-Ecosystem/cli/mineru-open-api@latest

Quick Start

# Parse a local HTML file (requires token)
mineru-open-api extract page.html -o ./out/

# Parse a remote HTML URL (requires token)
mineru-open-api extract https://example.com/page.html -o ./out/

# Parse a live web page (requires token)
mineru-open-api crawl https://example.com/article -o ./out/

Authentication

Token required:

mineru-open-api auth             # Interactive token setup
export MINERU_TOKEN="your-token" # Or via environment variable

Create token at: https://mineru.net/apiManage/token

Capabilities

• Supported input: local .html file or remote HTML URL
• HTML requires extract or crawl (token required)
• HTML is NOT supported by flash-extract
• Language hint with --language (default: ch, use en for English)

Notes

• HTML is NOT supported by flash-extract — use extract or crawl
• For live web pages with dynamic content, use crawl instead of extract
• Output goes to stdout by default; use -o <dir> to save to a file or directory
• All progress/status messages go to stderr; document content goes to stdout
• MinerU is open-source by OpenDataLab (Shanghai AI Lab): https://github.com/opendatalab/MinerU