Pricing

Rating

Developer

Actor stats

Categories

Share

Document Structure Extractor

Turn Markdown documents into structured JSON — heading tree, section text, code blocks, links, and parsed tables. Pure parsing, deterministic, no LLM cost.

What it does

For each input document it extracts:

• Title (first # heading) and preamble text
• Nested section tree: level, heading, body text, character counts, children — fenced code blocks never miscounted as headings
• Code blocks with language tags and line numbers
• Links ([text](url))
• Tables parsed into header + rows
• Stats: lines, characters, heading and code-block counts

Input

{
"documents":["# Guide\n\nIntro.\n\n## Setup\n\n```bash\npip install x\n```"]
}

Output (one dataset item per document)

{
"title":"Guide",
"sections":[
{
"level":1,"heading":"Guide","text":"Intro.",
"children":[{"level":2,"heading":"Setup","...":"..."}]
}
],
"code_blocks":[{"lang":"bash","code":"pip install x","line":7}],
"links":[],
"tables":[],
"stats":{"lines":9,"chars":52,"headings":2,"code_blocks":1}
}

Typical uses

• Building tables of contents / outlines for documentation sites
• Feeding section-level structure into RAG ingestion pipelines
• Auditing docs: section sizes, code-block coverage, dead-link candidates