AST

markdownToAst() returns a parsed JavaScript object, so no JSON.parse() is needed.

Each node is an object with one key naming its type.
ThematicBreak is the bare string "ThematicBreak".
Leaf text (raw, table cells) is unparsed inline Markdown.

import { markdownToAst } from "ironmark";
 
const ast = markdownToAst("# Hello\n\n- [x] done");
// {
//   Document: {
//     children: [
//       { Heading: { level: 1, raw: "Hello" } },
//       { List: { kind: { Bullet: 45 }, start: 0, tight: true, children: [
//         { ListItem: { checked: true, children: [{ Paragraph: { raw: "done" } }] } },
//       ] } },
//     ],
//   },
// }
 
for (const node of ast.Document.children) {
  if (typeof node !== "string" && "Heading" in node) console.log(node.Heading.raw);
}

The TypeScript types describe the full tagged union as AstNode. AST_SCHEMA_VERSION bumps on any breaking change to the shape; see Constants and utilities.

Round-trip to Markdown

astToMarkdown() turns an AST back into normalized Markdown:

import { markdownToAst, astToMarkdown } from "ironmark";
 
const ast = markdownToAst("**Hello**");
const md = astToMarkdown(ast);
// "**Hello**"

Use it to normalize Markdown, or to edit a document as data and write it back out. doc.markdown() does both steps in one call.

Headings and node counts

import { markdownToAst, extractHeadings, summarizeAst } from "ironmark";
 
const ast = markdownToAst("# Hello\n\n## World\n\nA paragraph.");
 
extractHeadings(ast);
// [
//   { level: 1, text: "Hello", id: "hello" },
//   { level: 2, text: "World", id: "world" },
// ]
 
summarizeAst(ast);
// { blockCount: 3, nodeCounts: { Heading: 2, Paragraph: 1 } }

Edit this page on GitHub