The color conversion of NSAttributedString rely on the appearance. We should use the page's appearance in this case.
<rdar://problem/42632246>
Created attachment 345856 [details] Patch
Comment on attachment 345856 [details] Patch r=me, unofficial
Comment on attachment 345856 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=345856&action=review Is this testable in any way? > Source/WebCore/editing/cocoa/WebContentReaderCocoa.mm:128 > + auto* page = frame.page(); I was wondering if there's ever a case where page could be null, but it looks like we're okay here because createFragmentAndAddResources will bail if !frame.page().
(In reply to Wenson Hsieh from comment #4) > Comment on attachment 345856 [details] > Patch > > View in context: > https://bugs.webkit.org/attachment.cgi?id=345856&action=review > > Is this testable in any way? Not currently. I am working on dark mode testing support in general and have many tests to add.
Comment on attachment 345856 [details] Patch Clearing flags on attachment: 345856 Committed r234270: <https://trac.webkit.org/changeset/234270>
All reviewed patches have been landed. Closing bug.