A small unifiedJS plugin that turns an HTML/HAST tree into a LaTeX AST (see unified-latex) so you can render, lint, or emit .tex from HTML sources. Designed for book-like documents (ebooks, etc.) where you want headings, chapters, metadata, and inline formatting to map cleanly into LaTeX.
npm install hast-latex rehype-parse unified @unified-latex/unified-latex-util-to-stringThe plugin consumes a HAST tree (e.g., from rehype-parse) and returns a unified-latex AST you can stringify with @unified-latex/unified-latex-util-to-string.
import { unified } from 'unified'
import rehypeParse from 'rehype-parse'
import { hastLatex } from 'hast-latex'
import { unifiedLatexStringCompiler } from '@unified-latex/unified-latex-util-to-string'
const html = `<html><head><meta name="title" content="Example" /></head><body><h1>Hello</h1><p>World.</p></body></html>`
// Build HAST from HTML
const rehypeProcessor = unified()
.use(rehypeParse)
.use(hastLatex, { documentClass: 'book' })
const hast = rehypeProcessor.parse(html)
const latexAst = rehypeProcessor.runSync(hast)
// Turn LaTeX AST into a .tex string
const latex = unified().use(unifiedLatexStringCompiler).stringify(latexAst)
console.log(latex)Unified plugin that converts HAST → unified-latex AST.
Options:
documentClass:'article' | 'report' | 'book'(default:'book'). Used for the emitted\documentclass{...}macro.makeTitle:boolean(default:false). Whentrue, inserts\maketitleand uses metadata from the HTML<head>(e.g.,<meta name="title">,<meta name="author">,dc.title,dc.creator) to populate\title{}and\author{}.macroReplacements:Record<string, string>mapping CSS selectors to LaTeX macro names for inline styling (e.g.,{ 'b,strong': 'textbf', 'i,em': 'textit' }). This lets you customize how inline HTML is converted to LaTeX commands.customMetaSelectors:Partial<Record<'title' | 'author', string | ((node: Hast.Node) => Hast.Node)>>(optional). Overrides the default algorithms for finding certain metadata from the HTML. Useful when the HTML uses non-standard tags or attributes for metadata. Can be a CSS selector string or a function that takes a HAST tree and returns a single HAST node.
Default macroReplacements:
{
'b,strong': 'textbf',
'i,em': 'textit',
u: 'underline',
's,strike,del': 'sout',
}import { unified } from 'unified'
import rehypeParse from 'rehype-parse'
import { hastLatex } from 'hast-latex'
import { unifiedLatexStringCompiler } from '@unified-latex/unified-latex-util-to-string'
const html = `<!doctype html><html><head>
<meta name="title" content="My Book" />
<meta name="author" content="Jane Doe" />
</head><body>
<h1 class="starred">Intro</h1>
<p><span class="smcap">Small Caps</span> and <u>underline</u>.</p>
</body></html>`
const processor = unified()
.use(rehypeParse)
.use(hastLatex, {
documentClass: 'book',
makeTitle: true,
macroReplacements: {
'span.smcap': 'textsc',
u: 'underline',
},
})
const hast = processor.parse(html)
const latexAst = processor.runSync(hast as any)
const latex = unified().use(unifiedLatexStringCompiler).stringify(latexAst)
console.log(latex)- Currently focused on book-like prose from ebook sources. Lists, images, tables, math, footnotes, etc., are not yet mapped.
- HTML must include a
<html><head>...</head><body>...</body></html>structure for metadata extraction. - Output is a LaTeX AST; you choose when/how to stringify or compile it.
- API and output shape will change; pinned at
0.xwhile iterating.