Speakr is a personal, self-hosted web application designed for transcribing audio recordings
Python
/**
* Safe markdown rendering.
*
* marked.parse() passes raw inline HTML through, and its output is assigned to
* v-html sinks (chat/Inquire answers, admin banner/disclaimers). That content
* originates from LLM output and admin settings, so an embedded
* `<img src=x onerror=...>` would execute. Every such path must go through
* this helper, which sanitizes the parsed HTML with DOMPurify.
*
* Fails closed: if DOMPurify is somehow unavailable, the text is emitted
* escaped rather than as raw HTML.
*/
(function () {
function renderMarkdownSafe(text) {
if (text === null || text === undefined) return '';
var html;
try {
html = (window.marked && window.marked.parse)
? window.marked.parse(text)
: String(text);
} catch (e) {
html = String(text);
}
if (window.DOMPurify && typeof window.DOMPurify.sanitize === 'function') {
return window.DOMPurify.sanitize(html);
}
var div = document.createElement('div');
div.textContent = html;
return div.innerHTML;
}
window.renderMarkdownSafe = renderMarkdownSafe;
})();