@schlessera/brain-render-template

The shared document shell brain-kit wraps around content before it is rendered to PNG or PDF. Markdown in, a single self-contained HTML document out — inline CSS, system fonts, no external requests.

It is deliberately tiny and pure: marked and a stylesheet, no browser, no filesystem, no network. The rendering itself lives in @schlessera/brain-render-puppeteer.

Two callers share it, which is the point — a page shared from the app and a PDF produced on the command line are byte-identical for identical input:

• @schlessera/brain-ui-server → POST /api/render
• @schlessera/brain → brain render

Usage

import { buildHtmlDocument } from "@schlessera/brain-render-template";

const html = buildHtmlDocument({
  content: "# Trip plan\n\nDay one: arrive.",
  contentType: "markdown",
  title: "Trip plan",
});

| Option | Default | Meaning | |---|---|---| | content | — | Markdown source or a fragment of HTML | | contentType | — | "markdown" parses; "html" passes through | | title | "Shared from Brain" | <title>, escaped | | allowHosts | [] | Image hosts the renderer will resolve |

Remote images

The renderer denies the page network access, so a remote <img> would render as a broken-image box. Those are replaced with a visible placeholder instead:

<span class="remote-image">[Matterhorn at dawn — not embedded]</span>

data: URIs are always left alone. To let specific hosts through, pass allowHosts and give the renderer the same list — this option only stops the placeholdering; it does not grant the page any access on its own.

Print behaviour

The stylesheet includes a @media print block that keeps tables, blockquotes, code blocks and diagrams from splitting across pages, and keeps headings with the content that follows them.