Getting clean HTML out of a Google Doc

This trips people up first: Download → Web Page from the Doc's menu gives you a zip with a full HTML document and an embedded stylesheet — the whole page, styles and all.

What you want instead is the clipboard. Selecting content in a Doc and copying puts a rich-text HTML version on it, and that version carries the structure without the stylesheet.

Google's clipboard HTML puts a class on almost every element — c0, c1, c17, and lst-kix_ names on list items. They're generated per document and refer to CSS that stays behind in the Doc.

So they're not just noise. They're dead references: they do nothing on your page, they collide with class names of your own, and they make the markup unreadable when you go to edit it.

They're removed by pattern — c followed by digits, lst-kix_ anything, docs-internal-guid ids. Classes of your own that happen to be in the paste are left alone.

Docs wraps copied content in <b style="font-weight:normal">. It's a <b> tag that explicitly turns bold off — a quirk of how the editor tracks formatting internally.

Paste that anywhere the style attribute gets stripped, which is most CMSes, and the whole block turns bold. The tag is unwrapped here rather than kept, so the content comes out at the nesting level it should have been at all along.

The same pass removes empty <span> elements left behind after their class names go. Those are what make a two-paragraph paste twelve lines long.

Links in a Doc come out pointing at google.com/url?q=https://example.com/… rather than at the destination. Google uses it for click tracking inside the editor.

Published on a page, it means every outbound link on your site routes through Google, shows the wrong URL in the status bar, and breaks the day that redirector changes.

The wrapper is unwound back to the real destination, repeatedly if it's nested. Tracking parameters go too — utm_source and friends, gclid, fbclid, and a handful of others.

Headings, paragraphs, lists, tables, links, bold and italic — the document, in other words. Scripts, event handlers and inline styles are gone, because clipboard HTML can come from any page on the web and gets treated as untrusted input.

Two things the cleaner leaves for you. Google's empty spacer paragraphs come through as <p></p>, so delete the ones you don't want. And Docs' images are hosted on Google's servers with URLs that expire — download them and host them yourself, or they'll vanish from your page later.