TypeScript
Install the ESM htomd package for Node and use its typed API.
Requires Node 24 or later. The package is ESM and has no runtime dependencies.
Install
npm install @jamiedavenport/htomdQuickstart
import { convert, extract } from "@jamiedavenport/htomd";
const html = "<article><h1>Tea</h1><p>Steep gently.</p></article>";
const markdown = convert(html);
const document = extract(html, { url: "https://example.org/tea" });
console.log(markdown); // "# Tea\n\nSteep gently.\n"
console.log(document.metadata.title); // "Tea"API
convert(html: string, options?: ExtractOptions): string;
extract(html: string, options?: ExtractOptions): Document;Both functions are synchronous. ExtractOptions has one optional url field,
accepting a string, null, or undefined. It supplies context without fetching.
An explicitly empty URL stays "" in metadata. Invalid HTML or URL argument
types throw TypeError.
The package exports the Document, Metadata, Diagnostics, and
ExtractOptions types:
| Result | Fields |
|---|---|
Document |
markdown: string, metadata: Metadata, diagnostics: Diagnostics |
Metadata |
title, author, description, language, publishedTime, url, canonicalUrl: each string | null |
Diagnostics |
strategy: "semantic", "scored", "fallback", or "none"; notes: readonly string[] |
Results, nested objects, and notes are readonly and frozen at runtime. Missing
metadata is null. The JavaScript API uses camelCase; the CLI’s JSON uses
published_time and canonical_url.
Native behavior
Relative URLs use Node’s WHATWG URL. Whitespace follows JavaScript’s Unicode
rules, list counters accept ASCII decimal digits, and JSON-LD uses JSON.parse.
These can differ from Python on edge cases; the
package README
records the details.
See shared behavior and limitations and the CLI.