dom2md

Turn the live DOM of a page into Markdown. Heuristic, dependency-free, and tiny — designed to run in the browser (as a <script> tag or a content script), not on a server.

It walks the DOM, collects leaf text blocks plus structural elements (tables, lists with nesting/numbering, <hr>, <audio>/<video>), drops boilerplate (nav/footer/aside, ads, chat/cookie widgets), and derives heading levels (#–######) from each heading's font size relative to the page's base font.

Install

npm install dom2md

Usage

ESM / CJS

import { dom2md } from 'dom2md';
// const { dom2md } = require('dom2md');  // CommonJS

const markdown = dom2md();                 // converts document.body
const partial  = dom2md({ root: document.querySelector('article') });

Browser <script> tag

The package ships a standalone build at dom2md/global (dist/dom2md.global.js). Include it and call the global dom2md():

<script src="node_modules/dom2md/dist/dom2md.global.js"></script>
<script>
  const md = dom2md();
</script>

(It also exposes window.dom2md.)

Content script (browser extension)

Because the standalone build defines window.dom2md, you can drop it into a content_scripts entry exactly like the original extension:

{
  "content_scripts": [
    { "matches": ["<all_urls>"], "js": ["dom2md.global.js", "content.js"] }
  ]
}

// content.js — dom2md() is now a global
button.onclick = () => {
  const markdown = dom2md();
  // ...
};

API

dom2md(options?)

Returns the page (or subtree) as a Markdown string.

interface Options {
  root?: Node | null;      // root to convert, defaults to document.body
  minTextLength?: number;  // minimum text length to keep a block, defaults to 2
}

How it works

1. Collect leaf text blocks (elements with text and no block-level children) and structural elements.
2. Filter hidden / zero-sized blocks and boilerplate (nav/footer/aside, ad/chat/cookie widgets by id/class).
3. Classify text blocks: headings (level from font size vs. the page's base font), quotes, code, paragraphs.
4. Render Markdown with inline formatting (bold/italic/code/links/images), escaping literal special characters.

Build

npm install
npm run build   # emits dist/ (esm + cjs + iife + types)

dist/ is committed so consumers can grab dist/dom2md.global.js directly from the package without building.

License

MIT