ContextQMD
Back to Plate

Decorated clipboard and selected-text helpers should strip render-only wrappers and FEFF

docs/solutions/logic-errors/2026-04-04-decorated-clipboard-and-selected-text-helpers-should-strip-render-only-wrappers-and-feff.md

53.0.62.2 KB
Original Source

Decorated clipboard and selected-text helpers should strip render-only wrappers and FEFF

What happened

After decorated-text and mark-placeholder browser proofs landed, two browser semantics still needed to be made honest:

  • copying decorated text should serialize fragment semantics, not visual decoration wrappers
  • selected-text reads across mark placeholders should not leak FEFF sentinels

What fixed it

Two changes closed the gap:

  1. slate-react-v2 Editable now owns the browser clipboard bridge through ClipboardBridge.setFragmentData(...) and ClipboardBridge.insertData(...)
  2. slate-browser get.selectedText() now strips FEFF from the DOM selection string before returning it

That made the browser proofs line up with actual editor semantics:

  • decorated copy writes the application/x-slate-v2-fragment payload plus clean HTML/plain text
  • rendered highlight wrappers do not leak into clipboard HTML
  • selecting through a mark placeholder still reads semantic text instead of "a\uFEFF"

Why this works

Decorations and mark placeholders are renderer details.

They can and should affect what the user sees, but they are not the fragment semantics the editor should transport through copy/paste, and they are not real user text.

If the helper returns wrapper DOM or FEFF sentinel text, the helper is lying about editor semantics.

Reusable rule

For Slate browser helpers around rich text:

  • selected-text helpers should normalize out zero-width sentinels
  • clipboard helpers should prove fragment semantics, not renderer wrappers

If a copy payload contains decoration-only attrs or a selected-text read contains FEFF, the browser layer is leaking implementation detail instead of editor truth.