http://bugs.winehq.org/show_bug.cgi?id=7372
--- Comment #55 from Ken Thomases ken@codeweavers.com --- Just to be clear, the Windows CF_HTML format doesn't require _the_ complete document (the whole original HTML document from which the fragment was copied), but _a_ complete document (the fragment plus the start and end tags of the nodes in which it is contained). The link you gave gives examples of fragments copied from larger documents. By the way, this http://msdn.microsoft.com/en-us/library/windows/desktop/ms649015%28v=vs.85%29.aspx seems to be a more up-to-date article in terms of its descriptions of the context, although the examples have busted HTML.
We should surround the fragment with "<!--StartFragment-->" and "<!--EndFragment-->" since apps are also allowed to find it by searching for those strings, rather than using the offsets.