Please test everybody. I don't know LHelp well but I applied the patch and then tweaked some things because nobody else did.
I can tell Andrey Sobol's work is an improvement.
Does lHelp allow to read Russian text CHMs on Linux?
To be more precise, is Unicode fully supported? Cyrillic text is a good test indeed.
There was already an issue with THTMLWordHighlighter.CheckTextNode. See the bug report.
FPC's XML lib uses UTF-16 DOMString (UnicodeString) which must be handled correctly with our UTF-8 solution.
Please report any problems with Unicode handling.