Yes, so the HTML entities have to be decoded before being indexed. Let me also find out if there's a way to do this automatically within the docsearch scraper.